longs reads should have a timeout

[bluey]

If the computer sends a partial then the device should give up on the read at some point.

[bluey]

See about using watchdog timer at end of event loop.

[jdunn]

A precursory attempt at this is more annoying than it's worth. The watchdog timer cannot be disabled once it is enabled, and must be reset annoyingly often. For example, receiving 2 packets may take longer than the timeout. This basically means that the resets have to be liberally sprinkled through ALL code. Maybe the clock can be set for longer? Nor worth dealing with at this point since the problem this is designed to fix has never been seen in the wild.