Prioritize thread snapshot event processing to avoid zone initialization stalls

[bleggett]

Fixes: #45

Because zone snapshot events and syscall events are multiplexed over the same channel, depending on ordering and timings of zone startup, plugin init, and other things, it can be possible for syscall events to starve the event handling loop such that it never gets to the zone snapshot, and thus never properly initializes the zone watch.

This simply changes the logic to prefer processing zone initializing thread snapshots first, so they are always head-of-line, and not counted as part of the falco event limit (which is the correct behavior/what I should have done originally).

it also adds a few extra safeguards and logging fixes.

Tested with falco deployed via the Falco helm chart, versions:

Falco version: 0.43.0
Libs version:  0.23.1
Plugin API:    3.12.0
Engine:        0.58.0
Driver:
  API version:    8.0.0
  Schema version: 4.1.0
  Default driver: 9.1.0+driver