Update active stake cache to use control table

[rdlrt]

Description

Currently, active stake cache uses max(epoch_no) from existing tables to run delta cache on. But this may not always be accurate for pool and epoch, as during setup - values for epoch prior may not be available. Stake Snapshot Cache would provides data an epoch earlier to these tables, thus - relying on max(epoch_no) is no longer an accurate outcome.

Account's active stake cache does not require maintaining historical information and can be left as-is.

Motivation and context

Found recently that some of our instances reported empty array for pool_history endpoint. On verifying, found out epoch as well as pool cache tables could have missing entries due to our current mode of execution. This only started showing up recently on fresh instances restored via snapshot, wherein snapshot cache would fill entries prior to historical information is populated

How has this been tested?

On rdlrt2 instance

[hodlonaut]

Looks okay, so single control table entry is now in charge of keeping track of how the two different cache tables get refreshed/populated (my understanding). Just a small side comment, I find variable names a little bit confusing sometimes (i.e. if control table entry has a good descriptive one why don't we name the variable we extract its value into the same)