Remove file prefetching from FileStream

[Dandandan]

Is your feature request related to a problem or challenge?

Currently it is not effective

• It doesn't have any priority which to fetch first, it will always fetch a
• It applies the pruning filters directly (instead of delaying when starting processing)
• It makes it more likely index is loaded several times over if scanning the same file across threads.

Describe the solution you'd like

Remove the prefetching API based on next file, so we can implement IO pipelining better.

Describe alternatives you've considered

No response

Additional context

No response