[DON'T MERGE] feat(x/gov): add governors

[giunatale]

• You can only delegate to a single governor, and 100% of your governance VP or nothing, but you can always override the vote.
• Governors election is done on "estimated" VP exactly like validators, so only top X (configurable param) active governors by VP are counted for voting
• Governors can set their status between active and inactive but can change it only once every 4 weeks. Active governors can be elected (inactive governors are skipped) but the base account has automatically a self-delegation of all its VP to the governor and cannot redelegate. Inactive governors can re-delegate.
• Governors VP is "estimated" (I think it's an exact computation as I think I cover everything) using x/staking hooks so when slashing/bond/unbond/redel happen it triggers a VP update.
• It should be ok computationally and also ok in how data is stored for easy retrieval, however maybe not optimized for storage itself as there is the need for a couple of additional indexes. Need to also check how much the governor VP computation is accurate because it might allow to further simplify VP computation (skip part of it) in tally
• I also added an invariant but I am not sure how that impacts performance
• Currently missing all tests, code is also itself untested, it was a weekend project. No guarantees on quality or if it works at all (or as intended).

Based on discussions as recorded in https://gist.github.com/giunatale/95e9b43f6e265ba32b29e2769f7b8a37

EDIT:

• Governors need to have at least minSelfDelegation (param) tokens bonded in x/staking to be elegible for tally (active governors always require to "self-delegate" the governance VP from their base account)
• Added another invariant on governors delegations consistency with cumulative virtual validator shares stored

[giunatale]

An alternative system that does not make use of the aggregate VP is also available at:

• https://github.com/giunatale/atomone/tree/governors-no-aggregate-vp
• x/gov: remove aggregate VP for governors giunatale/atomone#1 (for a diff with this branch)

In brief:

• It removes the aggregate VP (and related indexing) for governors, so no need to loop over all involved governors in the slashing hook (which is removed here)
• Governor election is simplified to happen only among governors that voted, so if a governor had a lot of delegations but did not vote it basically gave up its spot. Actually since there is no need to limit the number of governors with this system the max governors param is removed. Only account for active governors that voted and meet the minimum self delegation requirement.
• Since there is no aggregate VP number, the optimization for quorum is not possible

[github-actions]

This pull request has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[arlai-mk]

Hi, thanks for your work on this, it really starts to look like the idea I tried to work on in March/April 2023 here (except I called governors "representatives"): https://github.com/arlai-mk/cosmos-sdk/tree/arlai/gov-complex-delegation

May I ask why the need to restrict to one governor only, with 100% of your VP, and not ability to spread your VP across multiple governors.
My original idea was to spread by percentage: you say that you want to delegate your Governance VP to X governors, with let's say 40% to governor #1, 30% to governor #2, and 30% to governor #3. This way, for each staking/slashing event, you can still recalculate the VP that is delegated to each governor.
I reckon it slightly complexifies the computations, but as you will need to recalculate anyway, I don't think the overhead is too big for a much better UX (ability to delegate to multiple governors).

Thank you.

[giunatale]

Thank you for your question. The reason is tally performance. With such a simplification the impact on tally is limited while complexity would potentially add another term if we allow multiple delegations. Look at https://gist.github.com/giunatale/95e9b43f6e265ba32b29e2769f7b8a37, and the comments because there are corrections along the way. Happy to explain more if needed.

Also, ultimately in practice there is no need to really spread your governance delegations if they are independent from staking and are purely related to governance, and you can freely redelegate at any times and override with your vote. For staking, spreading your delegations is important cause it helps distribute stake and achieve a higher nakamoto coefficient. But looking at the governance system independently, why would you need to have more than one representative? Other delegated governance systems (e.g. Optimism) also don't have this. And if you think about governance delegations as purely a system to facilitate governance, I doubt that adding multi-delegations really adds much. In the end, if you have a nuanced opinion, then you should vote yourself. The delegations/governor system is there to help people that don't have time/knowledge and I think it's better to ask them to find one governor that they like instead of a pletora.

Also, I invite you to actually look at https://github.com/giunatale/atomone/tree/governors-no-aggregate-vp because imho is the better approach (for more info check the comment from me above) and is more complete with tests and a bunch of fixes.

EDIT: btw the percentage is also what I had originally in mind. But again the tally complexity that way adds another term O(number of votes x validators x governors).