Improve Documentation

[sherrywang15]

Some good functionalities of the package are not sufficiently documented on https://pylift.readthedocs.io/, for example

• Most optional arguments are only documented in docstring
• The capability to initialize TransformedOutcome object with a tuple of custom training and testing dataframes (very neat flexibility) is not documented

[rsyi]

Adding the custom training test dataframes to the usage page in this PR: #29

Will spend some time parsing through the docs in the future to make a more comprehensive usage doc. :) Thanks for pointing this out.