fix(trie): Fixes to pruning of MerkleChangeSets

[mediocregopher]

Closes #18950

This PR contains a few related fixes:

• Adds MerkleChangeSets pruning segment to the default set (it wasn't previously present)

• Prune the StoragesTrieChangeSets table, not StoragesChangeSets

• Allows for configuration of the MerkleChangeSets PruneMode. The PruneMode must be set for this segment, as we never want to keep the full history of trie changesets. The default is Distance(10064). If someone really wanted to they could manually set it to Before(0) to keep all history.

• Because PruneModes now has a field which can't be None it doesn't make sense to have none or is_empty methods. These have been removed.

[shekhirin]

LGTM, did you test with an existing config.toml to ensure that nodes won't break when upgrading to a version with these prune config changes?

[mediocregopher]

LGTM, did you test with an existing config.toml to ensure that nodes won't break when upgrading to a version with these prune config changes?

Yup, tested locally with both empty reth.toml and a reth.toml having other pruning settings, it all worked fine 👍