Use polars for I/O

[endrebak]

Reading and parsing say GTF-files is likely to be a lot faster in polars. We could check if polars is installed on the users machine and use it to parse gtfs instead of pandas if that is the case.

[endrebak]

It should be an optional install in the pyproject.toml also.

[ghuls]

You could also try to use biobear for reading input formats efficiently:
https://github.com/wheretrue/biobear

[endrebak]

I tried using optimized polars code and reading gtfs were 5 seconds faster (30 seconds compared to 35 seconds)