fix(cadence-matching): stop tasklist when stopping the shardprocessor

[eleonoradgr]

What changed?
Unload the tasklist and stop it when the shard processor is stopped

Why?
When the shard is assigned to another instance we need to make sure the tasklist is stopped, otherwise the tasklist keep processing the old task that are also requested from the new instance and this causes "Persistent store operation failure".

How did you test it?
Deployed the fix to a deployed environment, where cadence-matching is onboarded to the SD and checked that there are no more conflict in the updates of the task

Potential risks
Potential risk is that task lists can be stopped, and since the tasklists stopped are the one with higher load this can cause more errors "Failed to complete task" due to the list shutdown.

Release notes

Documentation Changes