Further tweaks to bookloupe checks

[windymilla]

1. Should a colon following a footnote number or letter be identified by "Query punctuation after"? (e.g. Footnote A:). Mentioned here
2. Is "unspaced bracket" doing what is wanted (e.g. footnote anchors). Also here

[tangledhelix]

I'm not sure how valuable "unspaced bracket" is for footnote anchors, but there are other cases where it will identify unprocessed DP markup:

• Diacriticals: [=e] for ē
• Scripts not handled via character picker: [Cyrillic: **]
• RTL languages: [Hebrew: **], [Arabic: **]
• Not-quite-right proofer notes PP'er may have missed otherwise: [* only one asterisk]

[tangledhelix]

... about 60 seconds after I made that comment, I thought "should I have read the code first?"

Now I see that the code is looking for a bracket with letters on both sides of it, so I'll say it probably won't find the 2nd or 3rd items in my list above. It would still locate diacriticals, and potentially bad proofer notes.

[tangledhelix]

Footnote anchors seem like something we could code around. They're caught in a case like this:

and they lived happily[A] ever after   ---> matches y[A

Footnote anchors follow a predictable pattern, so they're perfect for regex. We could detect and ignore them by inserting a negative lookahead into the positive lookahead (I've marked the new bit below):

(?<=\p{Letter})[][}{)(](?=\p{Letter}(?![\]]))
                                    ^------^

The negative lookahead is limited to ] intentionally, to more accurately match the use-case of ignoring a footnote.

There are edge cases this could miss. Imagine a really nasty scanno passing all the rounds, like windm]l]. That would defeat this regex (as long as the bad bracketing pattern occurred at the end of a word... otherwise it would still be caught, e.g. mi[l]meter). Further regex tomfoolery, or a second pass in code, might fix that edge case too.

[windymilla]

Consider whether to permit etc., as a permissible double punctuation