[Feature] Extract snapshot-reader logic into a standalone library

[kanatti]

Currently, RFS has two core responsibilities:

1. Reading from snapshots
2. Writing to the target cluster

There are other potential use cases where only reading from snapshots is valuable.

For example, embedding the snapshot-reading logic into a Spark/EMR application could enable:

1. Running parity/data quality checks across clusters by taking snapshots from both and comparing them directly in Spark.
2. During production data issues, taking a snapshot of the affected indices before mitigation, and then using Spark to assess the impact by reading directly from the snapshot.

Extracting the snapshot-reader logic into a standalone library would unlock these types of use cases outside of RFS.

[peternied]

@kanatti Can you tell me more about the shape of this refactor is there a specific limitation that you want to be have addressed. What prevents you from consuming the RFS library as it is?

[kanatti]

@peternied Thats right, we could consume RFS library directly. This was more in a direction to keep responsibilities separated. But we are not blocked on consuming RFS, let me try that out and will get back to you if we see any issues.