fix(asr): do not call InputFinished for a never-ending stream

A live caption stream never ends: the boundary of an utterance is silence,
not end-of-input, so InputFinished (which means "no more audio ever") does
not apply. Drop it and rely on the trailing audio alone; keep a short
silence padding as a safety net for end-of-file cases. Verified that the
file path still emits the final character without InputFinished.

实时字幕的音频流永不结束，一句话的边界是"静音"而不是"输入结束"，所以
不该调用语义为"再也不会有音频"的 InputFinished。改为只补一小段静音兜底，
实测文件模式的结果依然完整。文档也一并更正。

src/core/sherpaasr.cpp

@@ -92,9 +92,10 @@ std::string SherpaAsr::streamFinish()
     const std::vector<float> silence(pad, 0.0f);
     SherpaOnnxOnlineStreamAcceptWaveform(d->stream, 16000, silence.data(), (int32_t) pad);
     d->drain();
-    // 通知不再有输入，把尾部一并解出来
-    SherpaOnnxOnlineStreamInputFinished(d->stream);
-    d->drain();
+    // 注意：这里**不调用 InputFinished**。
+    // 实时字幕的音频流是永不结束的，一句话的边界是"静音"而不是"输入结束"；
+    // InputFinished 的语义（再也不会有音频了）对文件才成立，对我们不适用。
+    // 只需要补一小段静音，让模型把最后一个字吐出来即可。
     return d->resultText();
 }