fix(coordinator): don't route hotkey edges to QA while dictation runs

handle_pressed_edge previously routed the dictation hotkey to the QA
panel whenever qa_state.panel_visible was true, regardless of whether a
main dictation session was already active. If the user opened the QA
panel mid-dictation (or while polishing/inserting), the next dictation-
hotkey edge was routed into begin_qa_session, which calls
Recorder::start a second time on the same mic device. cpal usually
rejects the second build_input_stream on macOS/Windows but the
dictation session keeps running with no UX path to stop it from the QA
panel; on Linux/PipeWire the second open can succeed and you get two
concurrent capture streams competing for the audio device.

Symmetrically, handle_released_edge swallowed Released entirely when
the panel was visible — fine when QA owned the press, but if dictation
owned the press, Hold-mode dictation could never end because the
Released edge was eaten.

Fix: read inner.state.phase alongside panel_visible. If a dictation
session is non-Idle (Starting / Listening / Processing / Inserting),
both edges go to dictation regardless of panel_visible. The QA panel
stays open, just doesn't capture this hotkey.

3.3.4 (same PR, same files): open_qa_panel always emit'd CapsuleState::Idle
to sweep stale Done residue, but if dictation was mid-flight that
clobbered the in-flight capsule (Recording bar, Polishing progress, the
brief Done toast). Guard the sweep on dictation phase == Idle.

Audit IDs 3.3.1 + 3.3.4 (CONFIRMED 高 + 中).

Test: 183/183 lib tests pass. Manual verification (open QA panel during
dictation, press dictation hotkey — should stop dictation, not begin QA;
also Hold-mode dictation while panel visible should still stop on
release) requires the running app, to be done after merge.

openless-all/app/src-tauri/src/coordinator/dictation.rs

@@ -12,8 +12,14 @@ pub(super) async fn handle_pressed_edge(inner: &Arc<Inner>) {
     let was_held = inner.hotkey_trigger_held.swap(true, Ordering::SeqCst);
     if !was_held {
         // 路由：QA 浮窗可见时，rightOption 边沿走 QA；否则走主听写。详见 issue #118 v2。
+        // 例外：dictation session 已经在跑（Starting / Listening / Processing / Inserting），
+        // 即使 QA 浮窗被打开了，这条边沿也必须先走 dictation。否则 begin_qa_session 会
+        // 第二次抢同一个麦克风 device —— 在 Linux/PipeWire 上甚至会成功打开两路捕获，
+        // dictation 的 recorder 没人停；在 macOS/Windows 上 cpal 会拒绝第二次 build_input_stream
+        // 但 dictation session 仍在跑、用户找不到从 QA 面板停掉它的入口。审计 3.3.1。
+        let dictation_active = !matches!(inner.state.lock().phase, SessionPhase::Idle);
         let panel_visible = inner.qa_state.lock().panel_visible;
-        if panel_visible {
+        if panel_visible && !dictation_active {
             handle_qa_option_edge(inner).await;
         } else {
             handle_pressed(inner).await;
@@ -48,8 +54,12 @@ pub(super) async fn handle_released_edge(inner: &Arc<Inner>) {
     let was_held = inner.hotkey_trigger_held.swap(false, Ordering::SeqCst);
     if was_held {
         // QA 浮窗可见时，Option 行为是 press-toggle（不分 hold/release），release 边沿忽略。
+        // 与 handle_pressed_edge 的路由对称：dictation session 在跑时 Pressed 已经被路由到
+        // dictation，那 Released 必须也路由到 dictation —— 否则 Hold 模式松开热键时
+        // end_session 不会触发，dictation 永远停不下来。审计 3.3.1。
+        let dictation_active = !matches!(inner.state.lock().phase, SessionPhase::Idle);
         let panel_visible = inner.qa_state.lock().panel_visible;
-        if panel_visible {
+        if panel_visible && !dictation_active {
             return;
         }
         handle_released(inner).await;

openless-all/app/src-tauri/src/coordinator/qa.rs

@@ -2,7 +2,7 @@ use std::sync::Arc;
 
 use tauri::Emitter;
 
-use crate::coordinator_state::{initial_session_id, SessionId};
+use crate::coordinator_state::{initial_session_id, SessionId, SessionPhase};
 use crate::selection::SelectionContext;
 use crate::types::CapsuleState;
 
@@ -86,9 +86,16 @@ pub(super) fn open_qa_panel(inner: &Arc<Inner>) {
         state.selection = None;
         state.front_app = capture_frontmost_app();
     }
-    // 先把胶囊清干净，避免主听写上一次 Done 状态残留的 message/insertedChars
-    // 在 QA Done 阶段被 capsule UI 错误复用（"已之一粘贴这个 0" 那种）。
-    emit_capsule(inner, CapsuleState::Idle, 0.0, 0, None, None);
+    // 主听写 phase 是 Idle 才需要 sweep capsule —— 这里的语义是清掉「上一次 dictation
+    // Done 状态残留」的 message / insertedChars，让 QA 自己的 capsule 状态从干净起跑
+    // （否则 capsule UI 会出现 "已粘贴这个 0" 之类把上一次 inserted_chars 错误复用的
+    // 显示）。但如果 dictation 当前正处于 Recording / Polishing / Inserting / Done toast
+    // 显示中，强行 emit Idle 会把用户没看完的反馈抹掉、或者把 Polishing 中的进度条
+    // 卡死。审计 3.3.4。
+    let dictation_idle = matches!(inner.state.lock().phase, SessionPhase::Idle);
+    if dictation_idle {
+        emit_capsule(inner, CapsuleState::Idle, 0.0, 0, None, None);
+    }
     if let Some(app) = inner.app.lock().clone() {
         crate::show_qa_window(&app, "idle");
         let _ = app.emit_to(