feat: 实现跨 API 格式自动转换功能

- 新增端点级 format_acceptance_config 配置，控制是否接受跨格式请求
- 重构 EndpointFormDialog 为卡片式布局，支持内联编辑和格式转换开关
- StreamProcessor 实现流式响应的跨格式转换，支持 OpenAI/Claude/Gemini 互转
- CacheAwareScheduler 按端点格式筛选候选，同格式优先于跨格式
- 健康度/熔断按 Provider 端点格式分桶，而非客户端请求格式
- 新增 format_conversion_total 和 format_conversion_duration_seconds 指标
- 新增全局配置 format_conversion_enabled 控制总开关
- Input 组件新增 size="sm" 尺寸选项

tests/api/handlers/base/test_stream_processor_format_conversion.py

@@ -0,0 +1,103 @@
+import json
+from typing import Any, Dict, Optional
+from unittest.mock import AsyncMock
+
+import pytest
+
+from src.api.handlers.base.response_parser import ParsedResponse, ResponseParser, StreamStats
+from src.api.handlers.base.stream_context import StreamContext
+from src.api.handlers.base.stream_processor import StreamProcessor
+from src.core.api_format import register_all_converters
+
+
+class DummyParser(ResponseParser):
+    def parse_sse_line(self, line: str, stats: StreamStats) -> Optional[Any]:  # noqa: ANN401
+        return None
+
+    def parse_response(self, response: Dict[str, Any], status_code: int) -> ParsedResponse:
+        return ParsedResponse(raw_response=response, status_code=status_code)
+
+    def extract_usage_from_response(self, response: Dict[str, Any]) -> Dict[str, int]:
+        return {}
+
+    def extract_text_content(self, response: Dict[str, Any]) -> str:
+        return ""
+
+
+async def _empty_async_iter():
+    if False:  # pragma: no cover
+        yield b""
+
+
+@pytest.mark.asyncio
+async def test_create_response_stream_converts_claude_to_openai() -> None:
+    register_all_converters()
+
+    ctx = StreamContext(model="test-model", api_format="OPENAI")
+    ctx.client_api_format = "OPENAI"
+    ctx.provider_api_format = "CLAUDE"
+    ctx.needs_conversion = True
+
+    processor = StreamProcessor(request_id="test-request", default_parser=DummyParser())
+
+    response_ctx = AsyncMock()
+    response_ctx.__aexit__ = AsyncMock(return_value=None)
+
+    http_client = AsyncMock()
+    http_client.aclose = AsyncMock(return_value=None)
+
+    message_start = {
+        "type": "message_start",
+        "message": {
+            "id": "msg_1",
+            "type": "message",
+            "role": "assistant",
+            "model": "claude-test",
+            "content": [],
+            "stop_reason": None,
+            "stop_sequence": None,
+        },
+    }
+    content_delta = {
+        "type": "content_block_delta",
+        "index": 0,
+        "delta": {"type": "text_delta", "text": "Hi"},
+    }
+
+    prefetched_chunks = [
+        b"event: message_start\n",
+        f"data: {json.dumps(message_start)}\n".encode("utf-8"),
+        b"\n",
+        f"data: {json.dumps(content_delta)}\n".encode("utf-8"),
+        b"\n",
+    ]
+
+    out = b"".join(
+        [
+            chunk
+            async for chunk in processor.create_response_stream(
+                ctx,
+                byte_iterator=_empty_async_iter(),
+                response_ctx=response_ctx,
+                http_client=http_client,
+                prefetched_chunks=prefetched_chunks,
+            )
+        ]
+    )
+
+    text = out.decode("utf-8")
+    assert "event:" not in text
+
+    events = []
+    for line in text.splitlines():
+        if line.startswith("data: "):
+            events.append(json.loads(line[6:]))
+
+    assert len(events) >= 2
+    assert any(e.get("object") == "chat.completion.chunk" for e in events)
+    assert any(
+        e.get("choices", [{}])[0].get("delta", {}).get("content") == "Hi"
+        for e in events
+        if isinstance(e, dict)
+    )
+