Swifter

* Significantly reduced core dependencies of swifter library. See https://github.com/jmcarpenter2/swifter/issues/219 for discussion
- Big shout out to PeterJCLaw for starting this discussion and contributions from xquyvu as well
* Removed deprecated `loffset` parameter
- Thanks to bnavigator for identifying this bug
* Updated README to be more readable for darkmode users
- Thank you to MemphisMeng for identifying this gap

* Add secondary fallback for series applies
* Code refactoring for simplicity

* Enable indexing after a groupby, e.g. `df.swifter.groupby(by)[key].apply(func)`
* Improve groupby apply progress bar
- Previously, the groupby apply progress bar only appeared after the data was distributed across the cores.
- Now, the groupby apply progress bar appears before the data is distributed for a more realistic reflection of how long it took
* Additional groupby apply code refactoring and optimizations, including removing the mutability of the data within `ray`

* Enable users to pass in `df.index` as the by parameter for the df.swifter.groupby(by).apply(func) command

* Enable users to `df.swifter.groupby.apply`, which requires a new package (ray) that now available as an extra_requires.
* To use groupby apply, install swifter as `pip install -U swifter[groupby]`
* All credit goes to user diditforlulz273 for writing the performant groupby apply code, that is now part of swifter!

* Enable users to `force_parallel` which immediately forces swifter to jump to using dask apply. This enables a simple interface for parallel processing, but disables swifter's algorithm to determine the fastest apply solution possible.