Create CSV benchmark generator

[mchav]

Create a CSV data generator script that will generate benchmark data. Specifically I would like to measure the library's performance on all operations given:

• A very wide dataset - for benchmarking row level operations
• A very long dataset - for benchmarking columnar operations
• A dataset of all strings - for benchmarking reading without type inference
• A dataset with that's mostly numbers - for benchmarking reading with type inference

This is useful to have as a script rather than a dataset since it allows me to tune my numbers to be appropriate to the execution environment.

Also as an aside it would be useful to run this benchmark against:

• polars (in-memory)
• pandas
• Frames
• dplyr
• dataframes.jl

Then compare performance.