I've pulled snapshots of Wikipedia and Wiktionary text for the languages below, and established a baseline analysis with our current config. These cover all the custom analysis chains, other custom config (like using the ICU tokenizer), and a number of different scripts.

Chinese, Dzongkha, English, Finnish, French, Gan, Greek, Hebrew, Italian, Japanese, Javanese, Mirandese, Polish, Russian, Rusyn, Serbian, Slovak, Swedish, Tibetan, Turkish, Ukrainian

First draft done. Full details on MediaWiki.

Summary: Esperanto is missing, Serbian is broken. :( Chinese surrogates have been fixed! The ICU tokenizer has been updated. My sample is so old that many changes we made last year show up.

Issues:

• Serbian: Elastic search reports that "extra-analysis-serbian / 6.5.4-SNAPSHOT" is installed, but when I try to reindex, I get an error: Creating index...⧼Unknown filter type [serbian_stemmer] for [scstemmer]⧽
• Esperanto isn't in my original samples because it didn't have custom processing at the time I took them. However, I noticed that there was no Esperanto plugin in the new ES 6 batch of plugins.

Thanks for catching this, here is an updated version of the deb package which should fix the issues you discovered:
https://people.wikimedia.org/~dcausse/wmf-elasticsearch-search-plugins_6.5.4-alpha6~stretch_all.deb

Everything looks good now. Serbian (et al.) and Esperanto are working as expected. Thanks, @dcausse!