feat: implement video compositor review recommendations (#136)

- Add debug assertions for Y-plane raw pointer aliasing in convert.rs
- Clarify AVX2 packus cross-lane comments in simd_x86_64.rs
- Shrink ConversionCache entries when resolution decreases (>2x)
- Add max-entry bound (64) to FONT_CACHE with eviction
- Validate decoded image dimensions against max_canvas_dimension
- Fix fragile sleep-based test in pixel_convert.rs
- Derive Copy for Rect and BlitRect, remove unnecessary clones
- Remove compositor/pixel_ops/mod.rs re-export shim
- Extract shared bench_utils module for generate_rgba_frame
- Use font.metrics() instead of rasterize() for measurement
- Migrate compositor_only and pixel_convert benchmarks to criterion

Signed-off-by: Devin AI <devin@cognition.ai>
Co-authored-by: bot_apk <apk@cognition.ai>
Co-authored-by: Staging-Devin AI <166158716+staging-devin-ai-integration[bot]@users.noreply.github.com>

Author: staging-devin-ai-integration[bot]

Cargo.lock

@@ -45,6 +45,8 @@ indexmap = { version = "2.13", features = ["serde"] }
 
 opentelemetry = "0.31.0"
 
+criterion = { version = "0.5", default-features = false }
+
 # Profile settings
 [profile.release]
 debug = 1 # Include line tables for better stack traces in profiling

Cargo.toml

@@ -11,6 +11,7 @@ keywords = ["audio", "pipeline", "engine", "streaming", "realtime"]
 categories = ["multimedia", "network-programming"]
 readme = "README.md"
 publish = false
+autobenches = false
 
 [dependencies]
 # Skit's core library
@@ -56,6 +57,7 @@ script = ["streamkit-nodes/script"]
 compositor = ["streamkit-nodes/compositor"]
 
 [dev-dependencies]
+criterion = { workspace = true }
 serde_json = { workspace = true }
 tracing-subscriber = { version = "0.3", features = ["env-filter"] }
 futures = { workspace = true }

crates/engine/Cargo.toml

@@ -0,0 +1,72 @@
+// SPDX-FileCopyrightText: © 2025 StreamKit Contributors
+//
+// SPDX-License-Identifier: MPL-2.0
+
+//! Shared utilities for benchmark binaries.
+
+#![allow(dead_code)] // Not every bench uses every helper.
+
+use streamkit_nodes::video::pixel_ops::{rgba8_to_i420_buf, rgba8_to_nv12_buf};
+
+/// Standard resolutions used across benchmark suites.
+pub const RESOLUTIONS: &[(u32, u32)] = &[(640, 480), (1280, 720), (1920, 1080)];
+
+/// Generate an RGBA8 color-bar frame (opaque, all alpha = 255).
+#[allow(
+    clippy::many_single_char_names,
+    clippy::cast_possible_truncation,
+    clippy::cast_sign_loss,
+    clippy::cast_precision_loss
+)]
+pub fn generate_rgba_frame(width: u32, height: u32) -> Vec<u8> {
+    let w = width as usize;
+    let h = height as usize;
+    let mut data = vec![0u8; w * h * 4];
+    let bar_colors: &[(u8, u8, u8)] = &[
+        (191, 191, 191), // white
+        (191, 191, 0),   // yellow
+        (0, 191, 191),   // cyan
+        (0, 191, 0),     // green
+        (191, 0, 191),   // magenta
+        (191, 0, 0),     // red
+        (0, 0, 191),     // blue
+    ];
+    for row in 0..h {
+        for col in 0..w {
+            let bar_idx = col * bar_colors.len() / w;
+            let (r, g, b) = bar_colors[bar_idx];
+            let off = (row * w + col) * 4;
+            data[off] = r;
+            data[off + 1] = g;
+            data[off + 2] = b;
+            data[off + 3] = 255;
+        }
+    }
+    data
+}
+
+/// Generate an I420 frame by converting an RGBA frame.
+pub fn generate_i420_frame(width: u32, height: u32) -> Vec<u8> {
+    let rgba = generate_rgba_frame(width, height);
+    let w = width as usize;
+    let h = height as usize;
+    let chroma_w = w.div_ceil(2);
+    let chroma_h = h.div_ceil(2);
+    let i420_size = w * h + 2 * chroma_w * chroma_h;
+    let mut i420 = vec![0u8; i420_size];
+    rgba8_to_i420_buf(&rgba, width, height, &mut i420);
+    i420
+}
+
+/// Generate an NV12 frame by converting an RGBA frame.
+pub fn generate_nv12_frame(width: u32, height: u32) -> Vec<u8> {
+    let rgba = generate_rgba_frame(width, height);
+    let w = width as usize;
+    let h = height as usize;
+    let chroma_w = w.div_ceil(2);
+    let chroma_h = h.div_ceil(2);
+    let nv12_size = w * h + chroma_w * 2 * chroma_h;
+    let mut nv12 = vec![0u8; nv12_size];
+    rgba8_to_nv12_buf(&rgba, width, height, &mut nv12);
+    nv12
+}