Add explicit presignature index assignment

[zhouwfang]

#193

[Bridgerz]

Looking good! Can we also get tests for:

• Move test for reassign_presig_indices (epoch transition reassignment)
• Rust test for epoch mismatch rejection in mpc_sign_withdrawal_tx
• Rust test for the multi-batch-jump edge case

[Bridgerz]

This may warant changing the pending_withdrawal_ids data structure to something like a LinkedTable.

[bmwill]

So i'm not a huge fan of this because it requires unbounded iteration that must terminate inside of a single sui transaction. These types of things can lead to deadlock situations where a txn can never terminate because it takes too much gas or loads too many objects. imo i don't think we need this, or the ability to reassign indexes in the end_reconfig function and instead should introduce another function that can be used to re-key an explicitly provided set of ids. Given the epoch is encoded in there we can iterate through these offchain and selectively bump them before we go to signing them.

Thoughts? @manolisliolios can keep me honest here too

[Bridgerz]

Good point. puts a lot of load on the end of epoch tx as well as possible deadlock on that critical tx. I think a new function for reassigning explicit withdrawals makes the most sense. We will need to guard this new function with a certificate like we do for each stage of the withdrawal flow (ex. confirm_withdrawal)

[zhouwfang]

My original approach was lazy reassignment at signing time. I turned to this approach because the original approach requires a fair amount of plumbing (new RPC endpoint, and leader logic to fan out, aggregate, submit), and there was some urgency to unblock devnet. But it may be no longer urgent to merge this PR now as the issue it tries to fix is very unlikely to happen (see my elaboration in chat), and I can turn back to the original approach. WDYT?