[async] Check if args/kwargs are JSON Serializable while running locally

[mariamrf]

Context

• When running an async @task locally, the args and kwargs are passed as they are (even if they're not JSON Serializable).
• It would be nice to have this check when running locally so we can test for and predict failures of that kind.

Expected Behavior

• When a @task is run locally with invalid args/kwargs (non-serializable), it should break/raise an exception as this is the actual behavior when it's deployed.

Actual Behavior

• When a @task is run locally with invalid args/kwargs (non-serializable), it just runs as a normal function and therefore, we can't catch errors of this kind in the tests.

Possible Fix

In async.py:

            lambda_function_name = lambda_function_name_arg or os.environ.get('AWS_LAMBDA_FUNCTION_NAME')
            aws_region = aws_region_arg or os.environ.get('AWS_REGION')

            if (service in ASYNC_CLASSES) and (lambda_function_name):
                send_result = ASYNC_CLASSES[service](lambda_function_name=lambda_function_name,
                                                     aws_region=aws_region,
                                                     capture_response=capture_response).send(task_path, args, kwargs)
                return send_result
            else:
                validate_json_serializable({'args': args, 'kwargs': kwargs})
                return func(*args, **kwargs)

and in utilities.py:

def validate_json_serializable(thing):
    try:
        json.dumps(thing)
    except (TypeError, OverflowError):
        # raise more human-readable error here

Steps to Reproduce

1. Write a function that accepts a non-serializable type (like an instance of a class)
2. Run this function locally
3. Function runs!
4. Deploy the code to Lambda
5. Function breaks

Your Environment

• Zappa version used: 0.45.1
• Operating System and Python version: 3.6
• The output of pip freeze:
• Link to your project (optional):
• Your zappa_settings.py: