fix: remove local_ppois_to_compare, refactor diverged_subgraphs

[hopeyen]

Description

Should be a better fix for the metrics

• local_ppois_to_compare was recorded at store and delete of local attestations but using the wrong values. The original intention for this metrics is to track the number of public POIs to compare, but this in theory is just the number of deployments being actively crosschecked. This PR deprecate this metrics as it doesn't represent true values of local state, in favor of the existing comparison table
• diverged_subgraphs was recorded at each comparison_result interval based on compared results. Some subgraphs have diverged state but considered notFound for certain intervals when they are waiting for the new block checkpoint. This PR adds a new function to grab the comparison results of a certain type from the persisted states, instead of relying on a particular comparison interval. This should improve stability of the metrics and better represent the true number of subgraphs with diverged results.

Issue link (if applicable)

Resolves #70

Checklist

• Are tests up-to-date with the new changes?
• Are docs up-to-date with the new changes? (Open PR on docs repo if necessary)

[chriswessels]

Very clear PR description, nice job!