R4R: F1 storage efficiency improvements

[cwgoes]

Closes #3303

Makes the following changes to F1 to improve storage efficiency:

• Commission is force-withdrawn when a validator is removed, and the store record is cleared.
• DelegatorStartingInfo is deleted when a delegation's rewards are withdrawn
• Slash events, historical rewards, and current rewards are cleared when a validator is removed (all delegations will have been withdrawn)
• Reference counts are kept on historical rewards records, and records are deleted once no references to them (possible delegations which would need to reference them when withdrawing) remain.

F1 should now have storage usage ~= O(D + S + V), where D is the number of current delegations to active validators, S is the number of historical slashes (since the beginning of time) of active validators, and V is the number of active validators (also, the incentives are in favor of withdrawing rewards and decreasing storage usage). This seems reasonable as we already store delegations & active validators and slashes are costly / rare. This storage usage is checked in a simulation invariant.

Because of the way this is implemented, I'm not sure we need to store the reference count at all, but I suggest we leave it in for now until we're sure about that and sure we don't need to reference historical records for any other reason. Ref #3344.

Standard checklist:

• Linked to github-issue with discussion and accepted design OR link to spec that describes this work.
• Wrote tests
• Updated relevant documentation (docs/)
• Added entries in PENDING.md with issue #
• rereviewed Files changed in the github PR explorer

---

For Admin Use:

• Added appropriate labels to PR (ex. wip, ready-for-review, docs)
• Reviewers Assigned
• Squashed all commits, uses message "Merge pull request #XYZ: [title]" (coding standards)

[codecov]

Codecov Report

Merging #3333 into develop will decrease coverage by <.01%.
The diff coverage is 0%.

@@            Coverage Diff             @@
##           develop   #3333      +/-   ##
==========================================
- Coverage     54.9%   54.9%   -0.01%     
==========================================
  Files          132     132              
  Lines         9680    9684       +4     
==========================================
+ Hits          5315    5317       +2     
- Misses        4028    4030       +2     
  Partials       337     337

[cwgoes]

To-do:

• Reference counting seems to basically work, needs better testcases and docs
• See if current rewards, slash events, and all historical rewards can be cleared when validator is removed (since no delegations should exist anymore)
• Calculate some rough numbers on state usage (should be O(num_total_delegations))
• Ideally testcases could ensure this (no state leaks)

[alexanderbez]

LGTM 👌 ++ on the simulation storage invariant.

[rigelrozanski]

I think that my biggest comment is I would like to see a paragraph in the "implementation" spec (even if it doesn't exist yet, a new temporary .md file would suffice) describing the reward referencing counting mechanism from a high level.

Otherwise, code looks clean - have some general questions, some minor suggestions within.

[cwgoes]

@rigelrozanski Added a Markdown doc describing the rationale behind reference counting - bd0ebc8.

[rigelrozanski]

⭐️ 🚁 💾 🦅