[FEA] Better handling of reading lots of small Parquet files

[tgravescs]

Is your feature request related to a problem? Please describe.
A clear and concise description of what the problem is. Ex. I wish the RAPIDS Accelerator for Apache Spark would [...]

Describe the solution you'd like
A clear and concise description of what you want to happen.

Describe alternatives you've considered
A clear and concise description of any alternative solutions or features you've considered.

Additional context
Add any other context, code examples, or references to existing implementations about the feature request here.

Currently the readers (parquet for example) performs badly when there are a lot of small files. We perform much better at a small number of large files. We should improve performance of reading small files.

[tgravescs]

note that one issue with this is Spark has a feature where you can ask what the filename is that you are reading. Perhaps we can recognize when that feature is being used.

[tgravescs]

Note, we need separate issues for Parquet, Orc, etc.. all the ones we want to handle. I am concentrating on Parquet first.