Resolve collection ID before performing any range scan operations

Currently the collection ID is resolved on a per-request basis. This can result in situations where results from a single scan can originate from more than one collection. This could happen if a collection was dropped and then immediately recreated with the same name.

To implement this, range_scan_create_options just needs to be created with a non-zero collection_id (after it is fetched) instead of setting the collection_name member

No changes will be needed in the wrappers (other than a core update), this happens internally in the C++ core