Pandas Load Converters

[ieaves]

@sbrugman could you describe the problems you ran into with the converters here? I can take a look later

[sbrugman]

A user-friendly of integrating visions in pandas is by using visions' type inference while continuing the pandas API with minimal changes. Pandas' read_csv functionality has the dtype and converters parameters that can be used for this purpose. Loading data directly with visions' type inference has the potential to speed up the complete process.

In pandas, converters and dtype force the usage of the Python engine, which is slower than the C engine, negating the possible performance gain.