Arrow-pd-parser

First Major Release

- Changed API for reader and writer to simplify entry point to readers and writers
- Reduced support for multiple readers currently only uses Pandas for CSV and JSONL and Arrow for Parquet. Codebase can be expanded in the future.
- Automatically reads from S3 or local depending on string filepath
- README.md discusses use of package. Changes are not backwards compatible.
- Pushed arrow dependency to at least version `5.0.0`. This is to reflect changes made in the parquet reader that is necessary for packages in our other eco systems.

- made it so partition data is not cast

- added warning for complex data types not being cast (no current support)

- Added new Exception class should make it easier to run try/except as users can catch the new Exception class (`PandasCastError`)

- Expanded parse module to `arrow_parser` and added a `pandas_parser` module. But import the previous function in the `__init__.py` of parse so that it is backwards compatible with previous releases.
- `pandas_parser` module added to give users option of arrow reader or pandas (for CSV and JSONL only).
- Added some basic tests for the pandas_parser

- Updated docs
- Added skipped test which demonstrates known error in package (to be addressed in future update)
- Removed default indent of 4 from jsonl writer
- Updated the dependency for pyarrow allowing v2 and v3
- Upgraded the dependency for Pandas