feat: Support multiple extras in requirements constraint advertisement

[timmc-edx]

Update the base template's setup.py to support the full syntax of package extras.

• Support commas and whitespace inside extras declarations, e.g. some_package[extra_one, extra_two]
• Raise exception if we spell packages differently in requirement and constraint files, including presence/absence of extras.

Merge checklist:
Check off if complete or not applicable:

• Changelog record added
• Documentation updated (not only docstrings)
• Fixup commits are squashed away
• Unit tests added/updated
• Manual testing instructions provided
• Noted any: Concerns, dependencies, deadlines, tickets

[timmc-edx]

I should figure out how to do automated testing of this. For now, some manual testing:

1. In event-bus-kafka, change setup.py to read from test.in rather than base.in (so that there's a package with extras in the requires)
2. From event-bus-kafka repo, run the cookiecutters update_setup_py.sh script. Read .git/cleanup-python-code-description and observe that confluent_kafka was without its extras in the old requires, but carries the extras in the new requires.
3. Add confluent_kafka<=2.3.0 to the constraints.txt file. Run python setup.py bdist_wheel and observe the error:

   Exception: Encountered both "confluent_kafka[avro,schema-registry]" and "confluent_kafka" in requirements and constraints files; please use just one or the other.

[rgraber]

can we update the comment to include how the regex treats [] ? it's a pretty hard regex to read

[timmc-edx]

How would you feel about using verbose mode? Not something I think I've ever used before, but I agree that this regex is... pretty dense.

    requirement_line_regex = re.compile(
        r"""
          (
            # The base name of the package
            [%s]+
            # Optionally, an extras section, as in `some-package[extra-one, extra-two]`
            (?:
              \[[%s,\s]+\]
            )?
          )

          # Optionally, a version constraint like `>=3.0`
          (
            [<>=][^#\s]+
          )?
        """
        % (re_package_name_base_chars, re_package_name_base_chars),
        re.VERBOSE
    )

Or even using a raw format-string, which is apparently a thing you can do? And maybe adding some spaces inside some of the more confusing character classes:

    requirement_line_regex = re.compile(
        fr"""
          (
            # The base name of the package
            [ {re_package_name_base_chars} ]+
            # Optionally, an extras section, as in `some-package[extra-one, extra-two]`
            (?:
              \[ [{re_package_name_base_chars},\s]+ \]
            )?
          )

          # Optionally, a version constraint like `>=3.0`
          (
            [<>=][^#\s]+
          )?
        """,
        re.VERBOSE
    )

[timmc-edx]

(I've pushed up a far simpler option. Verbose mode is cool but... probably overkill here.)

[rgraber]

A comment here with an example entry would be helpful. It took me a minute to figure out how it was being used.

[timmc-edx]

Sure, maybe something like this:

    # e.g. {"django": "Django", "openedx-events": "openedx_events"}

[rgraber]

Suggested change

	- Fail packaging if requirements are spelled differently in different places
	- Fail packaging if requirements are named differently in different places or have different extras added

[rgraber]

It's not really spelling.

Suggested change

	This ensures that packages are spelled consistently so we can match
	This ensures that packages are named consistently so we can match

[rgraber]

One suggestion for comment, otherwise LGTM

[rgraber]

Just to be super de dooper clear

Suggested change

	# e.g. {"django": "Django", "openedx-events": "openedx_events"}
	# e.g. {"django": "Django", "openedx-events": "openedx_events", "confluent-kafka":"confluent-kafka[avro]"}

[timmc-edx]

The quality checks will probably yell about line length. Maybe I'll use confluent_kafka[avro] to demonstrate both underscore and extras?