The analysis settings for wikibase may create a set of analyzers/token_filter/char_filters prefixed per language.
Currently it generates 1200+ analyzers and and most of them are identical.
Only analysis components like token and char filters are deduplicated.
Deduplicating analyzers is not entirely trivial as they are referenced from the mapping config builders and all of them are expected to be there.
It might perhaps make sense to quickly evaluate the perf gain of doing such optimization (possibly measuring index creation and node startup).
AC:
- the number of analyzers created on wikibase indices is largely reduced