Lint: Validate PEP fields against PEP 12

[CAM-Gerlach]

Adds some more cheap Pygrep checks to conservatively validate the formally-specified PEP fields against PEP 12 (i.e. the required fields are present, and non-empty fields match any required format), with a few exceptions:

• The Author/Sponsor/PEP-Delegate have a strictly-defined format in the spec, and I prototypes a regex to enforce it; however, due to numerous deviations in practice and the non-triviality of conforming them automatically, I don't attempt to enforce that here, and leave that to a future PR, if interest exists
• Similarly, it was straightforward to modify the Created regex to also check Post-History, but given the number of deviations that would need to be corrected, the non-triviality of doing so programmatically, and the likely lesser interest in parsing the data, I didn't go through with that for now
• The Discussions-To fields doesn't have a strictly defined format in PEP 12; in practice it conforms to either {URL}, {email_address}, or {name} <{URL or email_address> following the Author/etc field, but as this is complex to parse/validate and not formally specified, I left this out

This caught a number of minor issues, which I fixed:

• A couple leading zeros on PEP numbers, as prohibited by PEP 12
• Some isolated instances of Python-Version not following the specified X.Y(.Z) format: e.g. 2.6 / 3.0 instead of the standard 2.6, 3.0, 3000 instead of 3.0, N/A/blank on non-standards-track PEPs that are not required to have them, etc.
• A URL missing HTTPS, and one with an errant space
• A few rejected/differed/withdrawn PEPs that had out of date TBD or empty Resolved fields (though the validation allows this, so as to not break intentional use)
• Some misc typos in fields

In some cases, some of the fields had extra information in parenthesis after the validated value; I allowed this in the regexes, and didn't modify or remove any such instances; nor did I remove any fields with non-empty values just to conform to the specification.

Running the full suite on the entire repo adds around ≈10 extra seconds to the lint job, but as this is still far faster than the build job, it makes no significant difference in practice.