Sessions

[fominok]

As much as it is possible it's unfortunate to have data and proof diverged, especially when debugging some issue. To keep fetched data and proof we're investigating in sync we need a debugging session mechanism.

Since GroveDB already has cheap snapshots, we can use them to create a session, meaning all the fetched data and proofs will relate to the one exact state. And of course, the session can be discarded to open a new one to work with the newest state.

For the best experience, we can also check subtree hashes on creating a next session to unload obsolete ones