fix(azure_speech): 修复流式播放时speaking状态未正确更新的问题

确保在流式播放模式下speaking状态在适当的时候更新，避免状态不一致。同时优化了sessionid为空时的处理逻辑。

local_plugins/azure_speech/ios/azure_speech/Sources/azure_speech/AzureTtsHelper.swift

@@ -596,8 +596,10 @@ public class AzureTtsHelper: NSObject, ITtsService, AVAudioPlayerDelegate {
         
         synthesizer?.addSynthesisCompletedEventHandler { [weak self] _, _ in
             guard let self = self else { return }
-            self.speaking = false
             if self.suppressStreamPlayback { return }
+            if !self.isUsingPushStreamCapture {
+                self.speaking = false
+            }
             // 将当前合成的音频加入播放队列
             if !self.currentSynthesisBuffer.isEmpty {
                 os_log("调用链: 合成完成 入队播放 数据长度=%{public}dB session=%{public}@", log: self.log, type: .info, self.currentSynthesisBuffer.count, self.sessionid)
@@ -621,11 +623,13 @@ public class AzureTtsHelper: NSObject, ITtsService, AVAudioPlayerDelegate {
         
         synthesizer?.addSynthesisCanceledEventHandler { [weak self] _, event in
             guard let self = self else { return }
-            self.speaking = false
             
             // 减少待处理文本计数
             self.pendingTextCount = max(0, self.pendingTextCount - 1)
             if self.suppressStreamPlayback { return }
+            if !self.isUsingPushStreamCapture {
+                self.speaking = false
+            }
             
             if self.isUsingPushStreamCapture {
                 self.audioPlaybackQueue.async {
@@ -664,11 +668,11 @@ public class AzureTtsHelper: NSObject, ITtsService, AVAudioPlayerDelegate {
      */
     private func handleSynthesizedAudioChunk(_ data: Data) {
         if suppressStreamPlayback { return }
-        if !speaking { return }
+        if sessionid.isEmpty { return }
         notifyAudioData(data)
         guard isUsingPushStreamCapture else { return }
         audioPlaybackQueue.async {
-            if !self.speaking { return }
+            if self.sessionid.isEmpty { return }
             let pcm = self.stripWavHeaderIfNeeded(data)
             if !pcm.isEmpty {
                 self.pcmPendingData.append(pcm)
@@ -767,6 +771,7 @@ public class AzureTtsHelper: NSObject, ITtsService, AVAudioPlayerDelegate {
      */
 private func checkStreamPlaybackCompletedIfNeeded() {
         if activeStreamSynthesisCount == 0, pcmPendingData.isEmpty, scheduledBufferCount == 0 {
+            speaking = false
             hasNotifiedPlaybackStartedForStream = false
             os_log("调用链: 流式播放完成 session=%{public}@", log: log, type: .info, sessionid)
             notifyEvent(eventType: .playbackCompleted)