StateSync component refactor follow-ups

[tomyrd]

After #650 gets merged, a few follow-up issues should be tackled to improve the refactor:

• We should find a way to move the update_mmr_data function inside the StateSync process. It now feels out of place as it performs note record state transitions but is not part of the component.
• Address this comment about NoteUpdate. While we are at this, we should look into reducing the creation of NoteUpdates inside note_state_sync, most of them just contain one or two notes and shouldn't need a whole Vec.
• Address this comment about locking accounts. If we decide to keep the stale hash check inside the store, we should move it to their respective lock_account functions inside each store.
• Change the interface so the sync isn't performed as steps. Instead it should be done as completely in a single call > related comment
• Address this comment about PartialMmr. We should add a dedicated struct in the client that implements some ease of use methods.

[igamigo]

One other thing is that I think we should review the return values from the introduced callbacks. Seems like there is not much value on having on_nullifier_received return a full TransactionUpdates: Since the idea is that this can detect only discarded transactions, maybe returning just a list of discarded transactions might be fine (or maybe this should not even be part of the callback, since you probably always want this to happen anyway).