fix(test): stabilize retry_with_timeout scenario

Patch time.monotonic alongside time.sleep in the retry timeout step
so stop_after_delay cannot elapse due to wall-clock overhead on
busy CI runners. The fake monotonic clock advances by 1µs per call,
keeping elapsed time well under the 0.01s timeout while preserving
retry semantics.

Closes #200

features/steps/retry_patterns_steps.py

@@ -572,13 +572,24 @@ def step_verify_async_execute_attempts(context, attempts):
 
 @when("I apply retry with timeout of {max_attempts:d} attempts and {timeout} seconds")
 def step_apply_retry_with_timeout(context, max_attempts, timeout):
-    """Apply retry_with_timeout while stubbing out sleep delays."""
+    """Apply retry_with_timeout while stubbing out sleep and monotonic clock."""
     timeout_seconds = float(timeout)
     original_sleep = time.sleep
+    original_monotonic = time.monotonic
+
+    _fake_base = original_monotonic()
+    _call_counter = [0]
+
+    def _fake_monotonic():
+        _call_counter[0] += 1
+        return _fake_base + _call_counter[0] * 1e-6
+
     time.sleep = lambda _seconds: None
+    time.monotonic = _fake_monotonic
 
     def restore_sleep():
         time.sleep = original_sleep
+        time.monotonic = original_monotonic
 
     # Register cleanup first so it runs even if the step body fails.
     context.add_cleanup(restore_sleep)