feat: GPU-downscale + software H.264 encode pipeline (WIP)

Add SwEncState in avhw.rs: GPU pipeline using scale_vaapi to downscale
4K BGRA -> 2K NV12 on AMD iGPU, then software encode with libopenh264.

- import_dma_buf_to_vaapi: av_hwframe_map based DMA-BUF import
- SwEncState: GPU filter graph (scale_vaapi) + NV12->YUV420P + libopenh264
- state_portal.rs: integrated SwEncState, auto DRM device detection
- vaapi_import_bench.rs: CPU vs GPU pipeline benchmark
- sw_encode_bench.rs: software encode benchmark

Benchmark results: GPU pipeline ~91 FPS theoretical (10.95ms/frame)
vs CPU pipeline ~33 FPS (30.21ms/frame).

Known issue: only 1 frame encoded in production recording,
diagnostic STATS logging added to debug frame flow.

src/main.rs

@@ -9,15 +9,15 @@ use wayland_client::globals::registry_queue_init;
 use wayland_client::Connection;
 
 // 各功能模块声明
-mod args;            // 命令行参数解析
-mod avhw;            // 音视频硬件加速
-mod backend_detect;  // 截屏后端自动检测（wlroots vs Portal/PipeWire）
-mod cap_portal;      // XDG Portal 屏幕捕获
-mod cap_wlr_screencopy;  // wlroots wlr-screencopy 截屏协议
-mod fps_limit;       // 帧率限制器
-mod state;           // wlr-screencopy 后端的主状态机
-mod state_portal;    // Portal/PipeWire 后端的主状态机
-mod transform;       // 图像变换（旋转/翻转）
+mod args; // 命令行参数解析
+mod avhw; // 音视频硬件加速
+mod backend_detect; // 截屏后端自动检测（wlroots vs Portal/PipeWire）
+mod cap_portal; // XDG Portal 屏幕捕获
+mod cap_wlr_screencopy; // wlroots wlr-screencopy 截屏协议
+mod fps_limit; // 帧率限制器
+mod state; // wlr-screencopy 后端的主状态机
+mod state_portal; // Portal/PipeWire 后端的主状态机
+mod transform; // 图像变换（旋转/翻转）
 
 use crate::args::Args;
 use crate::cap_wlr_screencopy::CapWlrScreencopy;
@@ -65,12 +65,8 @@ fn main() -> Result<()> {
 
     // 根据检测结果进入对应的事件循环
     match backend {
-        crate::backend_detect::CaptureBackend::WlrScreencopy => {
-            run_wlr_screencopy(args)
-        }
-        crate::backend_detect::CaptureBackend::PortalPipeWire => {
-            run_portal_pipewire(args)
-        }
+        crate::backend_detect::CaptureBackend::WlrScreencopy => run_wlr_screencopy(args),
+        crate::backend_detect::CaptureBackend::PortalPipeWire => run_portal_pipewire(args),
     }
 }
 
@@ -130,7 +126,7 @@ fn run_wlr_screencopy(args: Args) -> Result<()> {
     {
         let mut pfd = libc::pollfd {
             fd: wayland_fd,
-            events: libc::POLLIN,  // 监听可读事件
+            events: libc::POLLIN, // 监听可读事件
             revents: 0,
         };
         // timeout=0 表示非阻塞，立即返回当前 fd 状态
@@ -160,8 +156,8 @@ fn run_wlr_screencopy(args: Args) -> Result<()> {
     // signal_hook_mio 将 Unix 信号转换为 fd 可读事件，
     // 这样信号也可以通过 epoll 统一监听，不需要单独的信号处理器
     let mut signals = signal_hook_mio::v1_0::Signals::new(&[
-        signal_hook::consts::SIGINT,   // Ctrl+C
-        signal_hook::consts::SIGTERM,  // kill 命令默认信号
+        signal_hook::consts::SIGINT,  // Ctrl+C
+        signal_hook::consts::SIGTERM, // kill 命令默认信号
     ])?;
     poll.registry()
         .register(&mut signals, TOKEN_QUIT, Interest::READABLE)?;
@@ -305,11 +301,8 @@ fn run_portal_pipewire(args: Args) -> Result<()> {
     // 只注册信号 fd，没有 Wayland fd
     // 所以 poll.poll 在这里只负责检测 SIGINT/SIGTERM
     // 实际的帧采集完全依赖 poll_and_encode 的轮询
-    poll.registry().register(
-        &mut signals,
-        mio::Token(1),
-        mio::Interest::READABLE,
-    )?;
+    poll.registry()
+        .register(&mut signals, mio::Token(1), mio::Interest::READABLE)?;
 
     // 主事件循环（超时 10ms，比 wlr-screencopy 更短，因为不依赖 Wayland fd 唤醒）
     // 10ms 超时的作用是让循环高频转动，以便及时处理 PipeWire 投递的帧