Skip AutoEvents if they are taking too long

If an AutoEvent is scheduled to run, and the previous invocation is still running, log an error and skip the new run.
The current implementation continues to spawn new threads into the get_handler, which is likely to worsen the situation on an overloaded machine.

Start by tracing the AutoEvent scheduling path into get_handler and identify where a new thread is spawned while the previous invocation is still running. Verify the behavior on an overloaded machine, then confirm that the overlapping run is logged as an error and skipped rather than creating another thread.

Written by the indexing model from the issue text.