Sensormond changes to support configurable time for logging threshold and better handling of fatal signals

[gregoryboudreau]

Description

Two quality of life improvements for sensormond:

1. Allow for the user to configure timeout for printing time log warning (keep default at 30)
2. Cascade stop events (from signal handler) into voltage and current updaters to allow for more graceful checking during the update process, current setup has 10 second timeout and given the updaters can take some time to run can result in SIGKILLs getting sent during operations.

Motivation and Context

With a lot of sensors, the logs can be filled with warnings about exceeding the time to read even if it is expected by the vendor. Additionally, with this longer time, the SIGTERM can not be checked in enough time before supervisorctl falls back to using a SIGKILL

How Has This Been Tested?

Tested on Cisco Smartswitch, time configuration now no longer results in logs being triggered every loop and tested killing w/ supervisorctl stop sensormond and no longer see any error logs from system that were being hit when SIGKILL interrupted reads from driver.

Additional Information (Optional)

[mssonicbld]

/azp run

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).