feat(rust): advance state in post commit

[ion-elgreco]

Description

We advance the state in the post commit now, so it's done in a single location as per suggestion from @Blajda here: #2391 (comment)

This PR also supersedes this one: #2280

Related Issue(s)

• fixes Successful writes return error when using concurrent writers #2279
• fixes Generic DeltaTable error: Version mismatch with new schema merge functionality in AWS S3 #2262

[Blajda]

@ion-elgreco I've looked a this quick a bit and feel like a deep refactor is required for this. IMO if the commit logic is going to be responsible for providing the committed then it must return return a DeltaTableState / Snapshot. What are the expected scenarios where that struct will be None?
I also think it will be worth having the commit code take ownership of the passed Snapshot so a clone can be avoided.
Let me know what you think.

[ion-elgreco]

@Blajda Yeah it felt a bit strange to not be able to take ownership of the snapshot.

While working on it I didn't understand why the incoming snapshot could be None until I hit the create path, which obviously has no table snapshot.

We should just return always a snapshot. If the incoming snapshot is None, we can just create a snapshot after committing and return that. Do you agree?

[Blajda]

LGTM! Thanks for making the changes.