refactor(tests): update concurrent tests to explicitly pass mode

Address PR feedback:
- Update existing exception tests to explicitly pass concurrent_invocation_mode='throw'
- Remove redundant test_agent_concurrent_invocation_mode_throw_raises_exception test
- Rely on test_agent_concurrent_invocation_mode_default_is_throw for default verification

Author: strands-agent

tests/strands/agent/test_agent.py

@@ -2231,7 +2231,7 @@ def test_agent_concurrent_call_raises_exception():
             {"role": "assistant", "content": [{"text": "world"}]},
         ]
     )
-    agent = Agent(model=model)
+    agent = Agent(model=model, concurrent_invocation_mode="throw")
 
     results = []
     errors = []
@@ -2282,7 +2282,7 @@ def test_agent_concurrent_structured_output_raises_exception():
             {"role": "assistant", "content": [{"text": "response2"}]},
         ],
     )
-    agent = Agent(model=model)
+    agent = Agent(model=model, concurrent_invocation_mode="throw")
 
     results = []
     errors = []
@@ -2362,53 +2362,6 @@ def invoke():
     assert len(results) == 2, f"Expected 2 successes, got {len(results)}"
 
 
-def test_agent_concurrent_invocation_mode_throw_raises_exception():
-    """Test that concurrent_invocation_mode='throw' maintains the default behavior."""
-    model = SyncEventMockedModel(
-        [
-            {"role": "assistant", "content": [{"text": "hello"}]},
-            {"role": "assistant", "content": [{"text": "world"}]},
-        ]
-    )
-    # Explicit throw mode should behave same as default
-    agent = Agent(model=model, concurrent_invocation_mode="throw")
-
-    results = []
-    errors = []
-    lock = threading.Lock()
-
-    def invoke():
-        try:
-            result = agent("test")
-            with lock:
-                results.append(result)
-        except ConcurrencyException as e:
-            with lock:
-                errors.append(e)
-
-    # Start first thread and wait for it to begin streaming
-    t1 = threading.Thread(target=invoke)
-    t1.start()
-    model.started_event.wait()  # Wait until first thread is in the model.stream()
-
-    # Start second thread while first is still running
-    t2 = threading.Thread(target=invoke)
-    t2.start()
-
-    # Give second thread time to attempt invocation and fail
-    t2.join(timeout=1.0)
-
-    # Now let first thread complete
-    model.proceed_event.set()
-    t1.join()
-    t2.join()
-
-    # One should succeed, one should raise ConcurrencyException
-    assert len(results) == 1, f"Expected 1 success, got {len(results)}"
-    assert len(errors) == 1, f"Expected 1 error, got {len(errors)}"
-    assert "concurrent" in str(errors[0]).lower() and "invocation" in str(errors[0]).lower()
-
-
 def test_agent_concurrent_invocation_mode_default_is_throw():
     """Test that the default concurrent_invocation_mode is 'throw'."""
     model = MockedModelProvider([{"role": "assistant", "content": [{"text": "hello"}]}])