[Enhancement]: optimize search_sorted_pk

[sunby]

Is there an existing issue for this?

• I have searched the existing issues

What would you like to be added?

When we search sorted pks, we do not have to iterate all chunks.
First filter out the chunk that the pk exists in and search it in this chunk.

Why is this needed?

No response

Anything else?

No response

[xiaofan-luan]

Good catch.

What is the main use case for this? Can we first binary search on chunk first row?
Maybe we should set a index structure for each chunk's first row so we can binary search with?