Improve default text analysis settings

[cmark]

This PR adds the following feature to the index analysis settings:

• Set split_on_numerics to false by default on word_splitter token filter, which is used by default to tokenize descriptions of concepts, code terms, any text that requires tokenized analyzer search. This will keep words like 2D, 3D, 4D as single terms in the index as opposed to two terms a 2 and D, so search for 2D won't show up any result unless you have used term prefix matching.
• Add the keyword as available analyzer for searches where the search term should be kept as is

Also, made some minor API improvements.