make .detect_irregular_wt work again

[jonathanbratt]

This function works by detecting the pattern "[[Category:English irregular" in the wikitext for a word.

This pattern is often not getting picked up because it's not in a recognized section. For example, "ridden" and "taken" are picking up the past participle template, but not finding the irregular flag which is at the bottom of the English section (technically in the "anagrams" section). This is causing them to be processed as "ride -ed" and "take -ed".

[jonathanbratt]

note that irregulars might not all be flagged this way, too. I also see this in some cases:

{{head|en|verb form|cat2=irregular past participles}}

[jonthegeek]

We can't find a case where this still happens (these two words are purposefully broken into "ride" "-en" and "take" "-en" now). If we find an example we'll come back to this.