feat: associating tagging index with tx hash

[benesjan]

In this PR I implement the following:

• when transaction witness generation starts (beginning of pxe.proveTx) we prune indexes without tx hash to get rid of the mess caused by simulations (this should fix this issue),
• once the tx construction is finished (end of pxe.proveTx) we associate the new indexes with the given tx hash and mark the tx hash with PENDING status in the TaggingDataProvider,
• the synchronizer then loads all these tx hashes upon sync, requests the node for status and updates the statuses of these indexes (and reverts the statuses on reorgs),
• the possible statuses are: PENDING, DROPPED, FINALIZED (note that whether the tx reverted is irrelevant for indexes)
• the call to privateGetNextAppTagAsSender oracle function in a tx returns tag that corresponds to (highest_tracked_index + 1),
• the highest_tracked_index is found over all known indexes, even over dropped transactions BUT its maximum value is capped to LAST_FINALIZED_INDEX + WINDOW_LENGTH. This means that if too many txs get dropped we would link the new tx with the dropped one. Given that this is quite unlikely to happen if we set the WINDOW_LENGTH large enough I think this is fine.
• we prune all the metadata about indexes that are before the LATEST_FINALIZED_INDEX to save db space.

[benesjan]

• feat: associating tagging index with tx hash #17415 👈 (View in Graphite)
• refactor!: cleaning up tagging oracles #17372
• next

This stack of pull requests is managed by Graphite. Learn more about stacking.