Considering whether or not to have Cadence return immediately after an error. Part of "turning the corner," starting a stack off the I/O event queue is so that an exception will not be caught by some clever error handler, like Cadence. Cadence like control flow, or ad hoc control flow, that tries to catch exceptions will make the mistake of catching an exception that is meant to unwind the stack. I've used `setImmediate` to ensure that `abend` will cause an exception to entirely unwind. Now I'm playing around with determinism, which requires synchronicity. When I'm replaying log entries, it is important that the responses are synchronous for each log entry. (Is it? Can't I just replay the log entries in an async loop?) Should I get rid of the reliance on `setImmediate`? Should I profile the use of `nextTick`?