Forward disable_parallel_tool_use through Anthropic client-tool passthrough (PR #5711)

Round 11 reviewer consensus (10/10): the `disable_parallel_tool_use`
translation added in round 11b reached the Anthropic-compat server-tool
GGUF loop but not the analogous client-tool passthrough branch. A
client sending `/v1/messages` with custom tools plus
`tool_choice: {"type":"auto","disable_parallel_tool_use":true}` therefore
took the passthrough branch with the opt-out silently dropped on the
llama-server `/v1/chat/completions` body.

Thread the translated `anthropic_parallel_tool_calls` value through
`_anthropic_passthrough_stream` and `_anthropic_passthrough_non_streaming`
into the shared `_build_passthrough_payload`, which already knows the
field. Test pins both helpers' signatures and that the field reaches
the body via the payload builder.

studio/backend/routes/inference.py

@@ -4575,6 +4575,7 @@ async def anthropic_messages(
                 repetition_penalty = repetition_penalty,
                 presence_penalty = presence_penalty,
                 tool_choice = openai_tool_choice,
+                parallel_tool_calls = anthropic_parallel_tool_calls,
                 session_id = payload.session_id,
                 cancel_id = payload.cancel_id,
             )
@@ -4593,6 +4594,7 @@ async def anthropic_messages(
             repetition_penalty = repetition_penalty,
             presence_penalty = presence_penalty,
             tool_choice = openai_tool_choice,
+            parallel_tool_calls = anthropic_parallel_tool_calls,
         )
 
     if server_tools:
@@ -5010,6 +5012,7 @@ async def _anthropic_passthrough_stream(
     repetition_penalty = None,
     presence_penalty = None,
     tool_choice = "auto",
+    parallel_tool_calls = None,
     session_id = None,
     cancel_id = None,
 ):
@@ -5028,6 +5031,7 @@ async def _anthropic_passthrough_stream(
         min_p = min_p,
         repetition_penalty = repetition_penalty,
         presence_penalty = presence_penalty,
+        parallel_tool_calls = parallel_tool_calls,
         tool_choice = tool_choice,
         backend_ctx = llama_backend.context_length,
     )
@@ -5162,6 +5166,7 @@ async def _anthropic_passthrough_non_streaming(
     repetition_penalty = None,
     presence_penalty = None,
     tool_choice = "auto",
+    parallel_tool_calls = None,
 ):
     """Non-streaming client-side pass-through."""
     target_url = f"{llama_backend.base_url}/v1/chat/completions"
@@ -5177,6 +5182,7 @@ async def _anthropic_passthrough_non_streaming(
         min_p = min_p,
         repetition_penalty = repetition_penalty,
         presence_penalty = presence_penalty,
+        parallel_tool_calls = parallel_tool_calls,
         tool_choice = tool_choice,
         backend_ctx = llama_backend.context_length,
     )

studio/backend/tests/test_sampling_params_routing.py

@@ -874,3 +874,40 @@ def test_local_anthropic_disable_parallel_tool_use_translation():
     assert _extract(None) is None
     assert _extract("auto") is None  # string form (non-dict) → no opinion
     assert _extract({"type": "auto", "disable_parallel_tool_use": "yes"}) is None
+
+
+def test_local_anthropic_passthrough_helpers_accept_parallel_tool_calls():
+    """The Anthropic-compat client-tool passthrough helpers
+    (`_anthropic_passthrough_stream` /
+    `_anthropic_passthrough_non_streaming`) must accept and forward
+    `parallel_tool_calls` through `_build_passthrough_payload` so the
+    `disable_parallel_tool_use` translation works on the client-tool
+    branch the same way it does on the server-tool loop. Verified by
+    introspecting the signatures and confirming the field reaches the
+    body via the shared payload builder."""
+    import inspect
+
+    from routes import inference as route_mod
+
+    for fn in (
+        route_mod._anthropic_passthrough_stream,
+        route_mod._anthropic_passthrough_non_streaming,
+    ):
+        params = inspect.signature(fn).parameters
+        assert "parallel_tool_calls" in params, (
+            f"{fn.__name__} must accept parallel_tool_calls so the "
+            "Anthropic disable_parallel_tool_use translation reaches "
+            "the llama-server body on the client-tool branch"
+        )
+
+    body = route_mod._build_passthrough_payload(
+        openai_messages = [{"role": "user", "content": "hi"}],
+        openai_tools = [{"type": "function", "function": {"name": "x"}}],
+        temperature = 0.7,
+        top_p = 0.95,
+        top_k = 20,
+        max_tokens = 64,
+        stream = True,
+        parallel_tool_calls = False,
+    )
+    assert body.get("parallel_tool_calls") is False, body