Only cache check point state if it hasn't existed in other cache

[terencechain]

Background:
There's 2 main group of state storages. One used for stategen and one used for process attestations.

1.) The stategen's storages consists of hot_state_cache, epoch_boundary_state_cache and db
2.) The process attestation's state storage consist of check_point_cache.

Why do we need both?
Because of skip slots! check_point_cache stores beacon state for the epoch views. If there's a skip slot at 64, 63 and 62. Check point cache will get state of 61 and process slots and epoch up to 64. The processed state of 64 will be stored with the root of 61 and epoch at 2.

How can we do better?
We can reuse some of the spaces between them. For example if there's already a state at 64, we don't need to store it in the check point cache. We already know it's in the stategen stroage and it can be retrieved from stategen storage

What does this PR do?
This PR checks whether there's already a state in the stategen storage's bucket before saving it to the check point cache

[codecov]

Codecov Report

Merging #7037 into master will decrease coverage by 0.38%.
The diff coverage is 33.33%.

@@            Coverage Diff             @@
##           master    #7037      +/-   ##
==========================================
- Coverage   62.55%   62.16%   -0.39%     
==========================================
  Files         401      401              
  Lines       31638    31279     -359     
==========================================
- Hits        19792    19446     -346     
+ Misses       9067     9066       -1     
+ Partials     2779     2767      -12