Document how strings should be indexed

String indexing is confusing and involves a choice; in many cases the data we have is atypical of elasticsearch use cases where there's lots of free-text involved. At a minimum i'd like to provide documentation for plugin authors explaining the options (analyzed/non-analyzed), what they mean both for indexing and search and when is appropriate to use them (for instance, it almost always makes sense to treat ID fields as non-analyzed). It may also be desirable to indicate in facet mapping whether a string field is indexed or not, since some queries are only available on analyzed data and others on non-analyzed.