Build out cudf functionality

[tylerjthomas9]

One possible approach is to take the Pandas.jl interface

[beckernick]

It might be interesting to see if https://rapids.ai/cudf-pandas/ could be useful here. Not sure how it might work/not work with pandas.jl

[tylerjthomas9]

It might be interesting to see if https://rapids.ai/cudf-pandas/ could be useful here. Not sure how it might work/not work with pandas.jl

Congratulations on releasing this easy to use Pandas acceleration. It is a great win for the community. I have already adjusted some processes at work to use it. I think it will work with Pandas.jl.

The main thing I am looking at is getting all the features in Pandas.jl to work with PythonCall.jl. Pandas.jl uses PyCall.jl, so it I have to make a few adjustments. It shouldn't be too bad, but I just haven't had time to finish it up (I could just use both python interfaces at once, but I want to avoid this mess).