feat: enhance content filtering, token usage tracking, and stream error handling in chat-stream modules

tests/compat/expected/sse_content_filter_status.json

@@ -0,0 +1,8 @@
+{
+  "parts": [],
+  "finished": true,
+  "new_type": "text",
+  "content_filter": true,
+  "output_tokens": 77,
+  "error_message": ""
+}

tests/compat/expected/sse_leaked_content_filter.json

@@ -0,0 +1,7 @@
+{
+  "parts": [
+    {"text": "正常输出", "type": "text"}
+  ],
+  "finished": false,
+  "new_type": "text"
+}

tests/compat/expected/token_cases.json

@@ -1,6 +1,8 @@
 {
   "cases": [
     {"name": "ascii_short", "tokens": 1},
+    {"name": "whitespace_only", "tokens": 1},
+    {"name": "newline_only", "tokens": 1},
     {"name": "cjk", "tokens": 3},
     {"name": "mixed", "tokens": 4}
   ]

tests/compat/fixtures/sse_chunks/content_filter_status.json

@@ -0,0 +1,11 @@
+{
+  "chunk": {
+    "p": "response",
+    "v": [
+      {"p": "status", "v": "CONTENT_FILTER"},
+      {"p": "accumulated_token_usage", "v": 77}
+    ]
+  },
+  "thinking_enabled": false,
+  "current_type": "text"
+}

tests/compat/fixtures/sse_chunks/leaked_content_filter.json

@@ -0,0 +1,8 @@
+{
+  "chunk": {
+    "p": "response/content",
+    "v": "正常输出CONTENT_FILTER你好，这个问题我暂时无法回答"
+  },
+  "thinking_enabled": false,
+  "current_type": "text"
+}

tests/compat/fixtures/token_cases.json

@@ -1,6 +1,8 @@
 {
   "cases": [
     {"name": "ascii_short", "text": "abcd"},
+    {"name": "whitespace_only", "text": "   "},
+    {"name": "newline_only", "text": "\n"},
     {"name": "cjk", "text": "你好世界"},
     {"name": "mixed", "text": "Hello 你好世界"}
   ]