Fix timeouts for invocations on the command line.

When we use the chalice app in tests, we override the timeout.  When we use the chalice app on lambda, the timeout is valid.

However, in the current version of chalice we're using, calling `get_remaining_time_in_millis()` throws an exception when not run in a lambda.  If we get something unexpected (0 or None), we default to sys.maxsize.

chalice/app.py

@@ -85,9 +85,11 @@ def wrapper(*args, **kwargs):
                 future = executor.submit(method, *args, **kwargs)
                 time_remaining_s = chalice_app._override_exptime_seconds  # type: typing.Optional[float]
                 if time_remaining_s is None:
-                    time_remaining_s = min(
-                        API_GATEWAY_TIMEOUT_SECONDS,
-                        chalice_app.lambda_context.get_remaining_time_in_millis() / 1000)
+                    try:
+                        lambda_timeout_millis = chalice_app.lambda_context.get_remaining_time_in_millis()
+                    except:
+                        lambda_timeout_millis = sys.maxsize
+                    time_remaining_s = min(API_GATEWAY_TIMEOUT_SECONDS, lambda_timeout_millis / 1000)
                     time_remaining_s = max(0.0, time_remaining_s - EXECUTION_TERMINATION_THRESHOLD_SECONDS)
 
                 try: