I liked the option to specify Analyzer for jena-text, as implemented in
JENA-654. But I'd like to use an analyzer that is otherwise like KeywordAnalyzer but case-insensitive, for use in an autocomplete/typeahead UI widget. Lucene doesn't include such an analyzer, but there are several implementations of the same idea, e.g. in neo4j  and stargate .
I created my own implementation of such an analyzer and added code to use it from the assembler. Patch attached.
This analyzer is now in a new package org.apache.jena.query.text.analyzer, in case other analyzers for jena-text will appear in the future. If you don't like the new package, the class can of course be moved to org.apache.jena.query.text.
I also added a test for case-insensitivity. To avoid lots of duplicate boilerplate code, I slightly modified and subclassed the existing test for KeywordAnalyzer.
I'd love to see this in the next version of jena-text and Fuseki. Of course I'll rework the patch if necessary. I can also tweak the web documentation to mention this analyzer.