feat: enable parallel root during persistence

[Rjected]

An attempt at letting us enable parallel state root while persistence is running. Unfortunately this does cause some large state root calculation times after pruning:

2024-11-08T01:16:13.274042Z DEBUG engine::tree: Calculating state root in parallel, parent found in memory parent_hash=0x62b7207e38f8ee326718cd84122e65a0c529c95c7c972da3031e16be1a521c84 historical=0xce0b5402a4891a163a6a7bfc7082b64479be4f869f658f768a0fb9cbb3b0ad54
2024-11-08T01:16:13.276804Z DEBUG trie::parallel_state_root: pre-calculating storage roots len=41
2024-11-08T01:16:13.291558Z DEBUG provider::storage_writer: Appended block data range=2634289..=2634292
// ... more logs
2024-11-08T01:16:13.668259Z DEBUG engine::persistence: Running pruner block_num=2634292
2024-11-08T01:16:13.668301Z DEBUG pruner: Pruner started tip_block_number=2634292
// ... more logs
2024-11-08T01:16:13.669352Z DEBUG pruner: Pruner finished tip_block_number=2634292 elapsed=1.049125ms deleted_entries=136 limiter=PruneLimiter { deleted_entries_limit: Some(PruneDeletedEntriesLimit { limit: 10000, deleted: 136 }), time_limit: Some(PruneTimeLimit { limit: 100ms, start: Instant { tv_sec: 1135547, tv_nsec: 839583333 } }) } output=PrunerOutput { progress: Finished, segments: [(Headers, SegmentOutput { progress: Finished, pruned: 21, checkpoint: Some(SegmentOutputCheckpoint { block_number: Some(2634292), tx_number: None }) }), (Transactions, SegmentOutput { progress: Finished, pruned: 115, checkpoint: Some(SegmentOutputCheckpoint { block_number: Some(2634292), tx_number: Some(97718838) }) }), (Receipts, SegmentOutput { progress: Finished, pruned: 0, checkpoint: Some(SegmentOutputCheckpoint { block_number: Some(2634292), tx_number: Some(97718838) }) })] } stats=[(Headers, 21, Finished), (Transactions, 115, Finished)]
// ... many commits
2024-11-08T01:16:17.394935Z DEBUG engine::tree: Calculated state root root_elapsed=4.120853875s block=NumHash { number: 2634303, hash: 0x89638859120c12b17c853c18862b59e9a8f0e05e8d726f2b07d40e58c615e098 }

Still debugging why this occurs, it seems to happen after (and is somehow blocked by) many many rw-transaction commits, which is unexpected, since we should only be committing once in persistence.