Elasticsearch: Allow "safe" custom analyzers for keyword fields

[mattweber]

Please allow setting custom analyzers that use the keyword tokenizer and safe token filters on keyword fields. This would be great so we can use doc_values for things like case-insensitive sorting and icu collation without needing to create pipelines.

This was originally discussed in #12394 and the last comment from @jpountz said this can be done in a future 5.x release. This issue is to keep track of this potential enhancement.

[dakrone]

@mattweber I believe @jpountz is currently working on this!

[s1monw]

this is a duplicate of https://github.com/elastic/elasticsearch/issues/18064 and we are waiting to upgrade to lucene 6.2 see https://issues.apache.org/jira/browse/LUCENE-7355
closing this for now @mattweber please see the other issue

[mattweber]

Thanks @s1monw, did not see that.

[Omega359]

If @jpountz is working on #18064 can we have that issue updated to be assigned without the adoptme label? This feature would be a huge, huge deal for me to help reduce memory usage in my cluster.

[jpountz]

Done.