fix(agent): support Azure OpenAI gpt-5.x on chat/completions endpoint

Azure OpenAI exposes an OpenAI-compatible endpoint at
`{resource}.openai.azure.com/openai/v1` that accepts the standard
`openai` Python client. Two issues prevented gpt-5.x models from working:

1. `_max_tokens_param()` only sent `max_completion_tokens` for
   `api.openai.com` URLs. Azure also requires `max_completion_tokens`
   for gpt-5.x models.

2. The `codex_responses` upgrade gate unconditionally upgraded gpt-5.x
   to Responses API. Azure does NOT support the Responses API — it serves
   gpt-5.x on the regular `/chat/completions` path, causing a 404.

Fix: add `_is_azure_openai_url()` that matches `openai.azure.com` URLs.
- `_max_tokens_param()` now returns `max_completion_tokens` for Azure.
- The `codex_responses` upgrade gate skips Azure so gpt-5.x stays on
  `chat_completions` where Azure actually serves it.
- The fallback-provider api_mode picker also recognises Azure and stays
  on chat_completions.
- Tests cover max_tokens routing, api_mode behaviour, and URL detection.

gpt-4.x models on Azure are unaffected (already used chat_completions +
max_tokens, which Azure accepts for those models).

Salvage of PR #10086 — rewritten against current main where the
codex_responses upgrade gate gained copilot-acp / explicit-api_mode
exclusions.

tests/run_agent/test_run_agent.py

@@ -3386,6 +3386,61 @@ class TestMaxTokensParam:
         result = agent._max_tokens_param(4096)
         assert result == {"max_tokens": 4096}
 
+    def test_returns_max_completion_tokens_for_azure(self, agent):
+        """Azure OpenAI requires max_completion_tokens for gpt-5.x models."""
+        agent.base_url = "https://my-resource.openai.azure.com/openai/v1"
+        result = agent._max_tokens_param(4096)
+        assert result == {"max_completion_tokens": 4096}
+
+
+class TestAzureOpenAIRouting:
+    """Verify Azure OpenAI endpoints stay on chat_completions for gpt-5.x."""
+
+    def test_azure_gpt5_stays_on_chat_completions(self, agent):
+        """Azure serves gpt-5.x on /chat/completions — must not upgrade to codex_responses."""
+        agent.base_url = "https://my-resource.openai.azure.com/openai/v1"
+        agent.api_mode = "chat_completions"
+        agent.model = "gpt-5.4-mini"
+        # Mirror the routing logic from __init__
+        if (
+            agent.api_mode == "chat_completions"
+            and not agent._is_azure_openai_url()
+            and (
+                agent._is_direct_openai_url()
+                or agent._provider_model_requires_responses_api(
+                    agent.model, provider=agent.provider,
+                )
+            )
+        ):
+            agent.api_mode = "codex_responses"
+        assert agent.api_mode == "chat_completions"
+
+    def test_non_azure_gpt5_upgrades_to_codex_responses(self, agent):
+        """On api.openai.com, gpt-5.x must still upgrade to codex_responses."""
+        agent.base_url = "https://api.openai.com/v1"
+        agent.api_mode = "chat_completions"
+        agent.model = "gpt-5.4-mini"
+        if (
+            agent.api_mode == "chat_completions"
+            and not agent._is_azure_openai_url()
+            and (
+                agent._is_direct_openai_url()
+                or agent._provider_model_requires_responses_api(
+                    agent.model, provider=agent.provider,
+                )
+            )
+        ):
+            agent.api_mode = "codex_responses"
+        assert agent.api_mode == "codex_responses"
+
+    def test_is_azure_openai_url_detection(self, agent):
+        assert agent._is_azure_openai_url("https://foo.openai.azure.com/openai/v1") is True
+        assert agent._is_azure_openai_url("https://api.openai.com/v1") is False
+        assert agent._is_azure_openai_url("https://openrouter.ai/api/v1") is False
+        # Path-embedded azure string should still detect — we're ~substring matching
+        agent.base_url = "https://my-resource.openai.azure.com/openai/v1"
+        assert agent._is_azure_openai_url() is True
+
 
 # ---------------------------------------------------------------------------
 # System prompt stability for prompt caching