Optimize retrieval and caching

[anomit]

Is your feature request related to a problem?
The cacher was written as a quick drop in to pull certain values required by different components. But, it follows a stupidly simple design of repeated polling for the same values without any consideration for difference in states etc.

This is unnecessary and puts major load on the underlying RPC.

Describe the solution you'd like

• Fetch all slot/node information only once on startup
  • Subsequent updates can follow from event log updates and a minor range of backtracking on block ranges from the current head for error correction.
  • identify data market states that don't change, do not poll for them.

Describe alternatives you've considered
There is no alternative. This has to be done.

Additional context
NA

[chetna-mittal]

Completed implementation including separating static and dynamic state variables along with adding a flag to poll for static state variables too (for devnet/staging environments).

Implemented fetching blocks and processing events mechanism, have to include the offset fetching part and test the application.