fix: gracefully handle missing persisted_trie_updates

[klkvr]

Closes #13940

This PR divides ExecutedBlock into ExecutedBlockWithTrieUpdates and ExecutedBlock. in-memory state only contains ExecutedBlockWithTrieUpdates without tracking the persisted_trie_updates separately.

NewCanonicalChain::Reorg variant now only contains a Vec<ExecutedBlock> for reorg chain, thus not requiring us to obtain TrieUpdates for those.

Reorged blocks are no longer re-inserted into in-memory state

[mattsse]

I'm a bit torn about this because this could be very confusing

and not really a fan of introducing another type here.

but having the trienodes separate from the actual block makes a ton of sense and I have a feeling that this will be beneficial in the future.

leaning towards we should make this change
wdyt @rkrasiuk @Rjected

[mattsse]

this change kinda grew on me,

one caveat here is that this now always excludes trieupdates for reorged blocks so if users are currently relying on that this will no longer be possible.
but I don't think this is the case or is even something that is useful?

[klkvr]

but I don't think this is the case or is even something that is useful?

yeah I would expect that you wouldn't need this info generally, gvien that TrieUpdates are only useful when applying blocks to db, @rkrasiuk wdyt?