test: lock retry replacement semantics

Add regression coverage for gateway and CLI /retry behavior so retried messages replace the original user turn instead of accumulating duplicate user entries in history.

tests/test_cli_retry.py

@@ -0,0 +1,49 @@
+"""Regression tests for CLI /retry history replacement semantics."""
+
+from tests.test_cli_init import _make_cli
+
+
+def test_retry_last_truncates_history_before_requeueing_message():
+    cli = _make_cli()
+    cli.conversation_history = [
+        {"role": "user", "content": "first"},
+        {"role": "assistant", "content": "one"},
+        {"role": "user", "content": "retry me"},
+        {"role": "assistant", "content": "old answer"},
+    ]
+
+    retry_msg = cli.retry_last()
+
+    assert retry_msg == "retry me"
+    assert cli.conversation_history == [
+        {"role": "user", "content": "first"},
+        {"role": "assistant", "content": "one"},
+    ]
+
+    cli.conversation_history.append({"role": "user", "content": retry_msg})
+    cli.conversation_history.append({"role": "assistant", "content": "new answer"})
+
+    assert [m["content"] for m in cli.conversation_history if m["role"] == "user"] == [
+        "first",
+        "retry me",
+    ]
+
+
+def test_process_command_retry_requeues_original_message_not_retry_command():
+    cli = _make_cli()
+    queued = []
+
+    class _Queue:
+        def put(self, value):
+            queued.append(value)
+
+    cli._pending_input = _Queue()
+    cli.conversation_history = [
+        {"role": "user", "content": "retry me"},
+        {"role": "assistant", "content": "old answer"},
+    ]
+
+    cli.process_command("/retry")
+
+    assert queued == ["retry me"]
+    assert cli.conversation_history == []