WIP: ENH: Add script to migrate setup.py to pyproject.toml

[thewtex]

Contains a few assumptions, but will generate a good starting point for most
cases.

[jhlegarreta]

Thanks for doing this Matt 💯.

Gave it a try in InsightSoftwareConsortium/ITKTubeTK#162:

• The long_description is not parsed or is parsed in lieu of description and in the latter case multi-line descriptions are not handled correctly. The former should correspond now to the readme field I guess, so besides fixing the bug maybe printing a note that it is replaced by the value of readme could be helpful.
• The raw strings (i.e. starting with an r) in setup.py are not parsed correctly and make the script exit without success.
• The dependencies are not parsed correctly (in this case, maybe influenced by the raw strings and maybe due to each dependency being on a separate line and that introducing whitespaces).
• Keywords are not being parsed.
• Do we want to add Repository and Issues to the cookiecutter template and be also parsed here?
• Does authors introduce unnecessary whitespaces before name and before the closing bracket after email?
• Should all keywords be made lowercase automatically?

[jhlegarreta]

We will probably need additional steps to add the documentation dependencies, e.g.
https://github.com/InsightSoftwareConsortium/ITKTubeTK/blob/42104e631e8ef88d36144332a8bcae4070ab0fa7/docs/requirements.txt#L3

Also, we may need to warn users to add any additional dev or testing dependencies to [dev] or [test] etc. (e.g. "pre-commi"t, etc.).

[jhlegarreta]

Eventually, all README files should be translated to either markdown or restructuredtext 🙃.

[thewtex]

Agreed. We could do this as a follow-up step.

[thewtex]

@jhlegarreta thank you for the review and testing!

I will knock some of these out next time I exercise the script. Feel free to push to this branch, though, if you have fixes.