[Feature Request] skip entries by bytes in an iterator interface

[orarch]

Hello!

We have a single column family RocksDB with billion of entries.

We don't have any upfront knowledge of the key namespace.

We would like to be able to fast iterate over the DB and to be able to shard it into groups of around 100k keys.

Single threaded iteration is too slow for us.

Since we don't know the key namespace and structure, it would be beneficial to either have an efficient skipN API that skips N entries.

We don't mind if this API would be inaccurate (i.e we ask for N entries and it gives ~N).

Is there an existing way to get that?