pybigtools should allow setting autosql

[jackh726]

However, the autoSql still claims it's a bed3. Does autoSql have a way to declare at least something like "bed3+"?

Beyond that, it would be nice to able to store a custom autoSql definition, which we can use in bioframe.

Since in bioframe we can already infer common bed schemas from a dataframe and map numpy/pandas dtypes to autoSql types, we should be able to generate valid autoSql schemas from an input dataframe. Being able to store those upon writing a bigBed would keep that information in-band (and we could use it to automatically parse fields on read).

Originally posted by @nvictus in #54 (comment)

[jackh726]

Add an "autosql" option to BigBedWrite::write on master.

Realized that it might be nice to just overall provide more options when writing bbi files in python (the Rust API, and even CLI, offers a lot of options not currently exposed via the python API)