The system/config/db/site/data/collection.xconf file controls which lucene analyzer to use when indexing the data collection. It was set to use the WhitespaceAnalyzer. When I changed that to use the StandardAnalyzer instead and re-indexed the files, then the upper-case/lower-case issues went away.
I've done a bit of testing and the change does not seem to have introduced any problems, so I'm going to stick with it.
I also ran across the SnowballAnalyzer, which looks interesting, but I'll postpone investigating that until I get eXist working within Tomcat, as the arrangement with Jetty is annoying - particularly the implications for SVN.