Thanks for the story! Mocking seems the way for a predictable testing suite.
In relation to the live assertions, maybe I can still run the tests that interact with the sandbox environment in a separate test suite that is outside the golden path of deployments and, as you said, have some alert to check when things go south with the integration.
That would be my thought - two independent test suites.
We're a place where coders share, stay up-to-date and grow their careers.
We strive for transparency and don't collect excess data.