fix: Memoize use_table_data listener

[jnumainville]

Because the listener function was not memoized in use_table_data, the use_effect within use_table_listener was getting called unnecessarily, which means in some cases table updates can be missed during the start and stop of the listener.

Added fix and caution message in use_table_listener for now.

It might be worthwhile making use_table_listener more robust in cases where arguments changed, but the table specifically did not. This is not trivial though.

[jnumainville]

I was hoping this would fix those test issues, but I suppose not. Technically the check that is breaking isn't necessary but I want to figure out why it's breaking in this way.

[jnumainville]

Broken test was a multithreading issue. I've added some utilities to the hook testing for these cases.