Ability to automatically flush log before impending Lambda timeout

One of the frustrating things about running into a Lambda timeout when using structured logs is that nothing is logged. Ideally, when using `metric_scope`, it should be able to flush the log before the timeout expires using `context.get_remaining_time_in_millis()`, perhaps setting a property indicating this was the case (since there's a possibility the function actually does complete after the flush occurs and some logging was lost)