skip gpt-5-6 in e2e tests for now (#179)

* fix(tests/e2e): avoid /tmp for Codex home and skip gpt-5-6 Copilot models

- Codex refuses to create helper binaries under /tmp by design,
  so base the E2E codex home outside of pytest's tmp_path.
- gpt-5-6 models (luna, sol, terra) return 404 on mlflow/v1/chat/completions;
  add them to the Copilot skip list alongside the existing gpt-5.5 entry.

* update

* update

* add logs

---------

Co-authored-by: Lilly <lilly.gray@tecton.ai>

Author: lilly-luo

tests/test_e2e.py

@@ -404,6 +404,11 @@ class TestCodexLaunch:
     CODEX_INCOMPATIBLE_MODEL_FRAGMENTS = (
         # nano endpoint is unreliably slow and times out past the 60s budget.
         "gpt-5-4-nano",
+        # These endpoints are discoverable as responses-capable, but the
+        # backing OpenAI endpoint returns ENDPOINT_NOT_FOUND in the CI region.
+        "gpt-5-6-luna",
+        "gpt-5-6-sol",
+        "gpt-5-6-terra",
     )
 
     def _codex_models(self, e2e_state: dict) -> list[str]:
@@ -812,6 +817,8 @@ class TestCopilotLaunch:
         # gpt-5.5 rejects function tools + reasoning_effort on /chat/completions
         # ("Please use /v1/responses instead").
         "gpt-5-5",
+        # gpt-5.6 models similarly reject /chat/completions with 404.
+        "gpt-5-6",
     )
 
     def _all_models(self, e2e_state: dict) -> list[tuple[str, str]]: