Implement release_stake and release_unbonded for delegations

[ismellike]

closes #61

WIP - thinking in code for #128

[ismellike]

How can we handle rewards with delegator stake lazy calculation? There's going to be a state mismatch since we can't call update_stake for all users of a delegate that has released stake: https://github.com/Lay3rLabs/commitments/blob/297b7d0869fe4e9004ce22b4d875d98ec1bc703a/packages/apis/src/rewards/state.rs#L160-L167

[ueco-jb]

@ismellike

we can't call update_stake for all users of a delegate that has released stake

We could iterate over the keys of STAKED_AT using denom and delegate as prefixes and get a list of all of the stakers.
Or do you mean it's going to be too expensive for this unlucky user?

[ismellike]

@ismellike

we can't call update_stake for all users of a delegate that has released stake

We could iterate over the keys of STAKED_AT using denom and delegate as prefixes and get a list of all of the stakers. Or do you mean it's going to be too expensive for this unlucky user?

I meant every delegator needs their stake updated through the rewards calculator, because points_correction will not reflect the actual state. We cannot do this though, because we have an unbounded # of delegators per delegate.

Initial State (T0):
Delegate total_staked: 1000 tokens
points_per_share: 0
Stakers:
- A: stake 600, points_correction = 0, withdrawn_rewards = 0
- B: stake 400, points_correction = 0, withdrawn_rewards = 0

T1 (100 rewards distributed):
points_per_share = 100_000
Unclaimed rewards:
A: (600 * 100_000 - 0) / 1_000_000 = 60
B: (400 * 100_000 - 0) / 1_000_000 = 40

T2 (release_stake 500):
Total stake: 1000 -> 500
Without update_stake:
- A still recorded as 600
- B still recorded as 400
- points_corrections unchanged

T3 (C stakes 300):
Total: 500 -> 800
C: points_correction = 300 * 100_000 = 30_000_000

T4 (100 more rewards):
points_per_share = 225_000

Total unclaimed rewards would be:
A: (600 * 225_000 - 0) / 1_000_000 = 135 (60 from T1 + 75 from T3)
B: (400 * 225_000 - 0) / 1_000_000 = 90 (40 from T1 + 50 from T3)
C: (300 * 225_000 - 30_000_000) / 1_000_000 = 37.5 (just from T3)

Total rewards distributed would be: 262.5
But we only distributed 200 rewards (100 at T1 + 100 at T3)

[ismellike]

We also can't immediately normalize outflows, because this is defined over all denoms. addr -> outflows