Native support for reading compressed files

[jabraham17]

Feature Description
It would be nice if compressed files could be read directly by Arkouda. There is already some support for this for file types that have built-in compression (like parquet), but other file types can be compressed. My main example of this is a compressed csv file, which must be decompressed before ak.read_csv can be called.

In pure python+pandas, this can be cleanly done as follows

import pandas as pd
import bz2
with bz2.open("myCsv.bz2") as bzf:
  df = pd.read_csv(bzf)

This works because pd.read_csv takes either a filename or a file handle.

Example Behavior

My proposal for this would be something as follows

import arkouda as ak
data = ak.read_csv("myCsv.bz2", compression='bz2')

Here, compression (or decompress?) would inform the arkouda server about how it should do the decompression

Describe Workarounds

Currently, this is possible with a preprocessing script that either just decompresses the CSV or converts the compressed CSV to a compressed format that arkouda does support natively, like parquet.