studio: fix two backend CI test failures (capability dict + MTP recovery race) (#6464)

test_safetensors_capability_advertise: detect_reasoning_flags now returns a
reasoning_effort_levels key, so the none-template expectation must include it.

test_tensor_parallel::test_runtime_recovery_reloads_without_mtp: the assertion
raced the recovery thread, which sets _spec_fallback_reason just before its
finally clears _mtp_runtime_fallback_in_progress. Wait for the flag to clear
before asserting.

Co-authored-by: danielhanchen <michaelhan2050@gmail.com>

studio/backend/tests/test_safetensors_capability_advertise.py

@@ -107,6 +107,7 @@ def test_detect_safetensors_features_none_template_returns_all_false():
         "supports_reasoning": False,
         "reasoning_style": "enable_thinking",
         "reasoning_always_on": False,
+        "reasoning_effort_levels": [],
         "supports_preserve_thinking": False,
         "supports_tools": False,
     }

studio/backend/tests/test_tensor_parallel.py

@@ -365,6 +365,11 @@ def test_runtime_recovery_reloads_without_mtp(monkeypatch):
     while b._spec_fallback_reason != "runtime_error" and time.monotonic() < deadline:
         time.sleep(0.02)
     assert b._spec_fallback_reason == "runtime_error"
+    # The reload thread clears the single-flight flag in its finally, a beat after
+    # it sets the fallback reason -- wait for that instead of racing the thread.
+    deadline = time.monotonic() + 2
+    while b._mtp_runtime_fallback_in_progress and time.monotonic() < deadline:
+        time.sleep(0.02)
     assert b._mtp_runtime_fallback_in_progress is False