[coalesce] Implement specialized `push_batch_with_filter` for primitive array

[alamb]

Is your feature request related to a problem or challenge? Please describe what you are trying to do.
The idea here is that push_batch_with_filter currently calls filter which copies the passing values into a new array only to then copy the values again into the final output

If we can remove the intermediate result we may be able to improve performance substantially

Describe the solution you'd like

Describe alternatives you've considered

Additional context

[alamb]

I think this actually depends on

• [coalesce] Implement specialized BatchCoalescer::push_batch for PrimitiveArray #7763

[alamb]

I would like to continue this work but realistiically I don't have the focus at this time. Perhaps it is something that might interest @zhuqi-lucas as he is already starting to work on using this kernel downstream in DataFusion

[zhuqi-lucas]

Thank you @alamb for reminder, I’m interested and can take this on as part of my ongoing work！

I would like to continue this work but realistiically I don't have the focus at this time. Perhaps it is something that might interest @zhuqi-lucas as he is already starting to work on using this kernel downstream in DataFusion

[alamb]

Amazing -- thank you (for this any other things). I am sorry I can't keep up reviewing with your productivity!

[zhuqi-lucas]

Thanks so much, @alamb! I really appreciate the support — I’m excited to keep things moving and keep learning as I go.

Amazing -- thank you (for this any other things). I am sorry I can't keep up reviewing with your productivity!