Analyzer for query / terms normalisation

[agazzarini]

That would allow to reuse existing Lucene stuff for normalizing the query string before the quantities detection happens.
At this time, just a basic normalization (lowercasing) is done and it's hard-coded. Using an analyzer would allow a more powerful pre-analysis like diacritics, lowercasing, stemming (note that the current configuration requires you to provide singular / plural for the same unit)