feat: add display UI with splash screen, dashboard, and RSSI sparkline

Implement full SSD1306 OLED display support:
- Splash screen on boot (1.5s) with project name, version, board
- Dashboard showing radio state, config, counters, RSSI/SNR
- RSSI history sparkline (64-sample ring buffer, 2px-wide bars)
- DisplayOn/DisplayOff host commands routed via dedicated channel
- Pure render functions against generic DrawTarget (no hardware coupling)
- Display on/off control bypasses radio task entirely

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

vendor/hash32/src/fnv.rs

@@ -0,0 +1,37 @@
+use crate::Hasher as _;
+
+const BASIS: u32 = 0x811c9dc5;
+const PRIME: u32 = 0x1000193;
+
+/// 32-bit Fowler-Noll-Vo hasher
+pub struct Hasher {
+    state: u32,
+}
+
+impl Default for Hasher {
+    fn default() -> Self {
+        Hasher { state: BASIS }
+    }
+}
+
+impl crate::Hasher for Hasher {
+    #[inline]
+    fn finish32(&self) -> u32 {
+        self.state
+    }
+}
+
+impl core::hash::Hasher for Hasher {
+    #[inline]
+    fn write(&mut self, bytes: &[u8]) {
+        for byte in bytes {
+            self.state ^= u32::from(*byte);
+            self.state = self.state.wrapping_mul(PRIME);
+        }
+    }
+
+    #[inline]
+    fn finish(&self) -> u64 {
+        self.finish32().into()
+    }
+}

vendor/hash32/src/lib.rs

@@ -0,0 +1,141 @@
+//! 32-bit hashing algorithms
+//!
+//! # Why?
+//!
+//! Because 32-bit architectures are a thing (e.g. ARM Cortex-M) and you don't want your hashing
+//! function to pull in a bunch of slow 64-bit compiler intrinsics (software implementations of
+//! 64-bit operations).
+//!
+//! # Relationship to `core::hash`
+//!
+//! This crate extends [`core::hash`] with a 32-bit version of `Hasher`, which extends
+//! `core::hash::Hasher`. It requires that the hasher only performs 32-bit operations when computing
+//! the hash, and adds [`finish32`] to get the hasher's result as a `u32`. The standard `finish`
+//! method should just zero-extend this result.
+//!
+//! Since it extends `core::hash::Hasher`, `Hasher` can be used with any type which implements the
+//! standard `Hash` trait.
+//!
+//! This crate also adds a version of `BuildHasherDefault` with a const constructor, to work around
+//! the `core` version's lack of one.
+//!
+//! [`core::hash`]: https://doc.rust-lang.org/std/hash/index.html
+//! [`finish32`]: crate::Hasher::finish32
+//!
+//! # Hashers
+//!
+//! This crate provides implementations of the following 32-bit hashing algorithms:
+//!
+//! - [Fowler-Noll-Vo](struct.FnvHasher.html)
+//! - [MurmurHash3](struct.Murmur3Hasher.html)
+//!
+//! # Generic code
+//!
+//! In generic code, the trait bound `H: core::hash::Hasher` accepts *both* 64-bit hashers like
+//! `std::collections::hash_map::DefaultHasher`; and 32-bit hashers like the ones defined in this
+//! crate (`hash32::FnvHasher` and `hash32::Murmur3Hasher`)
+//!
+//! The trait bound `H: hash32::Hasher` is *more* restrictive as it only accepts 32-bit hashers.
+//!
+//! The `BuildHasherDefault<H>` type implements the `core::hash::BuildHasher` trait so it can
+//! construct both 32-bit and 64-bit hashers. To constrain the type to only produce 32-bit hasher
+//! you can add the trait bound `H::Hasher: hash32::Hasher`
+//!
+//! # MSRV
+//!
+//! This crate is guaranteed to compile on latest stable Rust. It *might* compile on older
+//! versions but that may change in any new patch release.
+
+#![deny(missing_docs)]
+#![deny(warnings)]
+#![no_std]
+
+extern crate byteorder;
+
+use core::fmt;
+use core::hash::BuildHasher;
+use core::marker::PhantomData;
+
+pub use fnv::Hasher as FnvHasher;
+pub use murmur3::Hasher as Murmur3Hasher;
+
+mod fnv;
+mod murmur3;
+
+/// A copy of [`core::hash::BuildHasherDefault`][0], but with a const constructor.
+///
+/// This will eventually be deprecated once the version in `core` becomes const-constructible
+/// (presumably using `const Default`).
+///
+/// [0]: https://doc.rust-lang.org/core/hash/struct.BuildHasherDefault.html
+pub struct BuildHasherDefault<H> {
+    _marker: PhantomData<H>,
+}
+
+impl<H> Default for BuildHasherDefault<H> {
+    fn default() -> Self {
+        BuildHasherDefault {
+            _marker: PhantomData,
+        }
+    }
+}
+
+impl<H> Clone for BuildHasherDefault<H> {
+    fn clone(&self) -> Self {
+        BuildHasherDefault::default()
+    }
+}
+
+impl<H> PartialEq for BuildHasherDefault<H> {
+    fn eq(&self, _other: &BuildHasherDefault<H>) -> bool {
+        true
+    }
+}
+
+impl<H> Eq for BuildHasherDefault<H> {}
+
+impl<H> fmt::Debug for BuildHasherDefault<H> {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        f.pad("BuildHasherDefault")
+    }
+}
+
+impl<H> BuildHasherDefault<H> {
+    /// `const` constructor
+    pub const fn new() -> Self {
+        BuildHasherDefault {
+            _marker: PhantomData,
+        }
+    }
+}
+
+impl<H> BuildHasher for BuildHasherDefault<H>
+where
+    H: Default + core::hash::Hasher,
+{
+    type Hasher = H;
+
+    fn build_hasher(&self) -> Self::Hasher {
+        H::default()
+    }
+}
+
+/// An extension of [core::hash::Hasher][0] for hashers which use 32 bits.
+///
+/// For hashers which implement this trait, the standard `finish` method should just return a
+/// zero-extended version of the result of `finish32`.
+///
+/// [0]: https://doc.rust-lang.org/core/hash/trait.Hasher.html
+///
+/// # Contract
+///
+/// Implementers of this trait must *not* perform any 64-bit (or 128-bit) operation while computing
+/// the hash.
+pub trait Hasher: core::hash::Hasher {
+    /// The equivalent of [`core::hash::Hasher.finish`][0] for 32-bit hashers.
+    ///
+    /// This returns the hash directly; `finish` zero-extends it to 64 bits for compatibility.
+    ///
+    /// [0]: https://doc.rust-lang.org/std/hash/trait.Hasher.html#tymethod.finish
+    fn finish32(&self) -> u32;
+}

vendor/hash32/src/murmur3.rs

@@ -0,0 +1,206 @@
+use core::slice;
+use core::mem::MaybeUninit;
+
+use byteorder::{ByteOrder, LE};
+
+use crate::Hasher as _;
+
+/// 32-bit MurmurHash3 hasher
+pub struct Hasher {
+    buf: Buffer,
+    index: Index,
+    processed: u32,
+    state: State,
+}
+
+struct State(u32);
+
+#[derive(Clone, Copy)]
+#[repr(align(4))]
+struct Buffer {
+    bytes: MaybeUninit<[u8; 4]>,
+}
+
+#[derive(Clone, Copy, PartialEq)]
+enum Index {
+    _0,
+    _1,
+    _2,
+    _3,
+}
+
+impl Index {
+    fn usize(&self) -> usize {
+        match *self {
+            Index::_0 => 0,
+            Index::_1 => 1,
+            Index::_2 => 2,
+            Index::_3 => 3,
+        }
+    }
+}
+
+impl From<usize> for Index {
+    fn from(x: usize) -> Self {
+        match x % 4 {
+            0 => Index::_0,
+            1 => Index::_1,
+            2 => Index::_2,
+            3 => Index::_3,
+            _ => unreachable!(),
+        }
+    }
+}
+
+impl Hasher {
+    fn push(&mut self, buf: &[u8]) {
+        let start = self.index.usize();
+        let len = buf.len();
+        // NOTE(unsafe) avoid calling `memcpy` on a 0-3 byte copy
+        // self.buf.bytes[start..start+len].copy_from(buf);
+        for i in 0..len {
+            unsafe {
+                *self.buf.bytes.assume_init_mut().get_unchecked_mut(start + i) = *buf.get_unchecked(i);
+            }
+        }
+        self.index = Index::from(start + len);
+    }
+}
+
+impl Default for Hasher {
+    #[allow(deprecated)]
+    fn default() -> Self {
+        Hasher {
+            buf: Buffer { bytes: MaybeUninit::uninit() },
+            index: Index::_0,
+            processed: 0,
+            state: State(0),
+        }
+    }
+}
+
+impl crate::Hasher for Hasher {
+    fn finish32(&self) -> u32 {
+        // tail
+        let mut state = match self.index {
+            Index::_3 => {
+                let mut block = 0;
+                unsafe {
+                    block ^= u32::from(self.buf.bytes.assume_init_ref()[2]) << 16;
+                    block ^= u32::from(self.buf.bytes.assume_init_ref()[1]) << 8;
+                    block ^= u32::from(self.buf.bytes.assume_init_ref()[0]);
+                }
+                self.state.0 ^ pre_mix(block)
+            }
+            Index::_2 => {
+                let mut block = 0;
+                unsafe {
+                    block ^= u32::from(self.buf.bytes.assume_init_ref()[1]) << 8;
+                    block ^= u32::from(self.buf.bytes.assume_init_ref()[0]);
+                }
+                self.state.0 ^ pre_mix(block)
+            }
+            Index::_1 => {
+                let mut block = 0;
+                unsafe {
+                    block ^= u32::from(self.buf.bytes.assume_init_ref()[0]);
+                }
+                self.state.0 ^ pre_mix(block)
+            }
+            Index::_0 => self.state.0,
+        };
+
+        // finalization mix
+        state ^= self.processed;
+        state ^= state >> 16;
+        state = state.wrapping_mul(0x85ebca6b);
+        state ^= state >> 13;
+        state = state.wrapping_mul(0xc2b2ae35);
+        state ^= state >> 16;
+
+        state
+    }
+}
+
+impl core::hash::Hasher for Hasher {
+    #[inline]
+    fn write(&mut self, bytes: &[u8]) {
+        let len = bytes.len();
+        self.processed += len as u32;
+
+        let body = if self.index == Index::_0 {
+            bytes
+        } else {
+            let index = self.index.usize();
+            if len + index >= 4 {
+                // we can complete a block using the data left in the buffer
+                // NOTE(unsafe) avoid panicking branch (`slice_index_len_fail`)
+                // let (head, body) = bytes.split_at(4 - index);
+                let mid = 4 - index;
+                let head = unsafe { slice::from_raw_parts(bytes.as_ptr(), mid) };
+                let body = unsafe {
+                    slice::from_raw_parts(bytes.as_ptr().offset(mid as isize), len - mid)
+                };
+
+                // NOTE(unsafe) avoid calling `memcpy` on a 0-3 byte copy
+                // self.buf.bytes[index..].copy_from_slice(head);
+                for i in 0..4 - index {
+                    unsafe {
+                        *self.buf.bytes.assume_init_mut().get_unchecked_mut(index + i) = *head.get_unchecked(i);
+                    }
+                }
+
+                self.index = Index::_0;
+
+                self.state.process_block(&self.buf.bytes);
+
+                body
+            } else {
+                bytes
+            }
+        };
+
+        for block in body.chunks(4) {
+            if block.len() == 4 {
+                self.state
+                    .process_block(unsafe { &*(block.as_ptr() as *const _) });
+            } else {
+                self.push(block);
+            }
+        }
+
+        // XXX is this faster?
+        // for block in body.exact_chunks(4) {
+        //     self.state
+        //         .process_block(unsafe { &*(block.as_ptr() as *const _) });
+        // }
+
+        // let tail = body.split_at(body.len() / 4 * 4).1;
+
+        // self.push(tail);
+    }
+
+    #[inline]
+    fn finish(&self) -> u64 {
+        self.finish32().into()
+    }
+}
+
+const C1: u32 = 0xcc9e2d51;
+const C2: u32 = 0x1b873593;
+const R1: u32 = 15;
+
+impl State {
+    fn process_block(&mut self, block: &MaybeUninit<[u8; 4]>) {
+        self.0 ^= pre_mix(LE::read_u32(unsafe { block.assume_init_ref() }));
+        self.0 = self.0.rotate_left(13);
+        self.0 = 5u32.wrapping_mul(self.0).wrapping_add(0xe6546b64);
+    }
+}
+
+fn pre_mix(mut block: u32) -> u32 {
+    block = block.wrapping_mul(C1);
+    block = block.rotate_left(R1);
+    block = block.wrapping_mul(C2);
+    block
+}