More compact API for fetching matching templates for a lexeme #221
Description
The Lexeme Forms gadget (https://www.wikidata.org/wiki/MediaWiki:Gadget-lexeme-forms.js) does a request to Lexeme Forms for every lexeme the user loads, e.g. on https://www.wikidata.org/wiki/Lexeme:L7, it will request https://lexeme-forms.toolforge.org/api/v1/match_template_to_lexeme/www/L7
The response from Lexeme Forms includes every template and whether or not it matches. The gadget ignores anything where the language or lexical category don't match, leaving just one template for L7.
Currently the response is around 68 kB, which compresses to about 4 kB. A JSON object with only the matching template would be around 134 bytes, uncompressed. 99% of the response is just thrown away, so including it every time is a waste of resources. The responses will also only get bigger over time, as more templates are added.
It would be better to have an API which only returns templates where the language and lexical category match (or at least where the language matches). The number of templates for a language will be limited, whereas Lexeme Forms is still a long way from covering all the languages Wikidata has lexemes for (and Wikidata is still a long way from covering all the languages with entries in a Wiktionary).