Add positionIncrementGap to keyword field

[JCRPaquin]

Why?

Keyword searches can span multiple independent values for a single document: if a document has “chemistry” and “mars” as two separate keywords then keyword:“chemistry mars” will match the document. This is not the expected behavior.

What?

Adds a new field type to the Solr schema: ads_text_gap. This type includes a position increment gap, which causes Solr to calculate token positions for separate field values as if they were separated by the gap value.

[kelockhart]

See my longer comment in-line, but basically, see if we can just incorporate positionIncrementGap in ads_text instead of creating a copy into ads_text_gap.

Also, this definitely needs some unittests.

[kelockhart]

This is a copy of ads_text, which is a little worrisome, as presumably these need to stay in sync. Is it ok to just set the positionIncrementGap for all ads_text fields instead of creating a new ads_text_gap field type? I'm guessing this only effects multiValued=true fields, which currently include title, comment, pubnote, all (what is this? do we use this?), caption, gpn, planetary_feature, and unfielded_search. I don't know about all and unfielded_search but for all the rest, this change should be neutral or positive (i.e. for gpn/planetary_feature, we will want to use ads_text_gap anyway, for the same reasons as for keyword)