Restoring a snapshot should just restore the documents that were visible when you took the snapshot.
Snapshots work by copying all of the files that make up the index into the repository and restoring them when you need them. They save space by not copying files that haven't changed since the last snapshot. They use reference counting to know when it is ok to delete the files. This works out well in an index that you are just adding documents to because most of the index files are write once. If you do lots of deletes or updates then the backups will be able to save less space.