Row group selection support in Parquet chunked reader

[karthikeyann]

          The chunked reader and row-group specification should work together fine (though I don't think we have a test specifically for it).  The only input to the chunked reader is "number of bytes in a chunk" so it doesn't care at all if the row groups being inspected in the first place are the entire file or part of it.

Originally posted by @nvdbaranec in #13348 (comment)

Chunked parquet reader ignores row group selection silently. Chunked parquet reader could support row group selection, and also predicate pushdown using row group filtering.

Enable user row group selection for chunked parquet reader.
As a follow up, add predicate pushdown support using row group filtering to chunked reader too.

[mhaseeb123]

@karthikeyann, do you think #15113 and #15764 together close this issue or is there any other work needed?

[karthikeyann]

Yes. These are enough.
Great work enabling chunking reader.