I probably won't develop this any further. For now the sqlite based SimpleSearch seems more efficient: https://github.com/kitsunet/Flowpack.SimpleSearch.ContentRepositoryAdaptor So that will get my attention. You should consider using that instead. I will keep this here for reference.
Please make sure you checked the Readme on https://github.com/kitsunet/Flowpack.ZendLucene about installation because it is a requirement for this.
Currently this package is rather experimental. It CAN already create a searchable index of all your nodes with the defaul configuration.
See the nodeindex:* commands on CLI for what you can do.
generally a:
./flow nodeindex:build
will generate the full index. After repeated builds you might want to run:
./flow nodeindex:optmize
Finally to search the index you can use
./flow nodeindex:find
Currently there is no Helper yet to access the index in Neos, but that will follow next. Also I didn't find good settings for tokenizing some fields yet. So some things are not optimal yet.