batched undelegations

[sampocs]

Context

Base PR for the batched undelegation work. This PR consists of only proto changes, but subsequent PR's will be merged into this before main.

Changelog

New Fields on HZU

Now that the StToken and NativeToken amounts are decremented in different places, I think it makes more sense to add explicit fields for each purpose. In this vein, I added the following:

NativeTokensToUnbond

• purpose is so that if we have failed batches, we know how much we have left that needs to be unbonded
• set right before the ICA is submitted
• decremented in the callback

StTokensToBurn

• purpose is to track tokens that need to be burned
• set right before ICA is submitted
• decremented in callback

ClaimableNativeTokens

• purpose is to track tokens that have been claimed
• set right before redemption sweep ICA is submitted
• decremented during each user claim

UndelegationTxsInProgress

• purpose is to prevent submitting a retry until we know all txs for a given epoch have been completed
• otherwise, it would could attempt to unbond with funds that were already unbonded (happens if the ack comes back after the epoch)
• This is incremented when each ICA is submitted and decremented in the callback

New Status Field

Additionally, we need to differentiate EUR's that are in a retry state vs those that are new unbondings. For this, I added the new status field UNBONDING_RETRY_QUEUE

Updated Callback Data

Finally, the callback data was updated to also store the number of stTokens. Given the fact that there's some precision error when breaking down a total stToken amount across validators, I think there are two options:

• Option 1: Only store native amount in callback, handle precision error by checking if the stAmount is below some epsilon, and if, so, rounding it down to 0
• Option 2: When building the ICA messages, capture the stToken remainder on the last validator and store it in the callback data

I opted for option 2 as it seems less error prone.

[riley-stride]

looks good, ty for the detailed description

[ethan-stride]

LGTM, thank you for explaining new fields.
As the github alerts point out, this change to the protos means that this PR alone has some inconsistent types now -- like in icacallback_undelegate on line 48 it still refers to the old field called SplitDelegations instead of the new SplitUndelegation but I assume that these are fixed in the later PRs which actually update the callbacks

[shellvish]

Looks great, thanks a lot for doing this! Will drop more questions if they come up as I go through the other PRs, but this looks great