test(streaming): match the exact chat completions route and type the chunks

Review raised two things worth taking. The route was matched by host alone,
so the fixture answered any method and path on the provider and the tests
would have stayed green if the request went somewhere else; it now matches
POST on the chat completions URL, and asserts the route was called. Pointing
that URL at a wrong path turns five of them red, so the match is doing work.

The SSE helpers took and returned bare dicts, which the repo's typing
conventions rule out and which let a malformed fixture through unnoticed.
They now carry TypedDicts for the chunk, choice, delta and usage shapes.

ReadOnly and TypedDict come from typing_extensions, matching the rest of the
codebase, because the unit shards run Python 3.12 and typing.ReadOnly landed
in 3.13.

tests/test_litellm/test_main.py

@@ -10,6 +10,7 @@ from fastapi.testclient import TestClient
 
 
 import urllib.parse
+from typing_extensions import ReadOnly, TypedDict
 from unittest.mock import MagicMock, patch
 
 import litellm
@@ -2959,15 +2960,40 @@ async def test_acompletion_resolves_provider_from_api_base():
     assert response.choices[0].message.content == "resolved"
 
 
-OPENAI_HOST = "api.openai.com"
+OPENAI_CHAT_COMPLETIONS_URL = "https://api.openai.com/v1/chat/completions"
 A_STREAMED_MODEL = "gpt-4o"
 
 
-def _sse(chunks: list[dict]) -> str:
+class _StreamedDelta(TypedDict, total=False):
+    content: ReadOnly[str]
+
+
+class _StreamedChoice(TypedDict):
+    index: ReadOnly[int]
+    delta: ReadOnly[_StreamedDelta]
+    finish_reason: ReadOnly[str | None]
+
+
+class _StreamedUsage(TypedDict):
+    prompt_tokens: ReadOnly[int]
+    completion_tokens: ReadOnly[int]
+    total_tokens: ReadOnly[int]
+
+
+class _StreamedChunk(TypedDict, total=False):
+    id: ReadOnly[str]
+    object: ReadOnly[str]
+    created: ReadOnly[int]
+    model: ReadOnly[str]
+    choices: ReadOnly[list[_StreamedChoice]]
+    usage: ReadOnly[_StreamedUsage]
+
+
+def _sse(chunks: list[_StreamedChunk]) -> str:
     return "".join(f"data: {json.dumps(chunk)}\n\n" for chunk in chunks) + "data: [DONE]\n\n"
 
 
-def _content_chunk(text: str, finish_reason: str | None = None) -> dict:
+def _content_chunk(text: str, finish_reason: str | None = None) -> _StreamedChunk:
     return {
         "id": "chatcmpl-stream",
         "object": "chat.completion.chunk",
@@ -2977,7 +3003,7 @@ def _content_chunk(text: str, finish_reason: str | None = None) -> dict:
     }
 
 
-def _usage_chunk(prompt_tokens: int, completion_tokens: int) -> dict:
+def _usage_chunk(prompt_tokens: int, completion_tokens: int) -> _StreamedChunk:
     return {
         "id": "chatcmpl-stream",
         "object": "chat.completion.chunk",
@@ -2993,7 +3019,7 @@ def _usage_chunk(prompt_tokens: int, completion_tokens: int) -> dict:
 
 
 def _a_stream(deltas: tuple[str, ...], prompt_tokens: int, completion_tokens: int) -> str:
-    body = [_content_chunk(delta) for delta in deltas[:-1]]
+    body: list[_StreamedChunk] = [_content_chunk(delta) for delta in deltas[:-1]]
     body.append(_content_chunk(deltas[-1], finish_reason="stop"))
     body.append(_usage_chunk(prompt_tokens, completion_tokens))
     return _sse(body)
@@ -3001,8 +3027,8 @@ def _a_stream(deltas: tuple[str, ...], prompt_tokens: int, completion_tokens: in
 
 @contextlib.contextmanager
 def _openai_streaming(body: str):
-    with respx.mock:
-        respx.route(host=OPENAI_HOST).mock(
+    with respx.mock(assert_all_called=True) as router:
+        router.post(OPENAI_CHAT_COMPLETIONS_URL).mock(
             return_value=httpx.Response(
                 200, text=body, headers={"content-type": "text/event-stream"}
             )