139190: javax.swing.text.BadLocationException: Invalid offset=907 not within <0, 892>

Fix 132306 - don't try to task scan huge files

Move HtmlFormatter parameter from completion request and analyzer scan request, and into the HTML getter calls where it belongs

GSF: Automatically handle cleanup of files deleted outside the IDE.

During startup, for each index, I already have a list of all the files known to the index.
(I load a special map of url to timestamp for each such file by using Lucene's support
to only load specified terms; this map is used during startup indexing to quicklycheck whether a file is new or has been updated outside of the IDE.).
Now, I also keep a list of all the files I've indexed that I also had a timestamp for (e.g. it
was known to the index). After indexing each source root, I check whether the size of these
two lists are identical. If not, files have been deleted outside of the IDE, and I diff
the two lists and delete the files that were in the index but no longer on disk.

(This was requested by PHP but is useful for all language clients since you can now be confident
that all matches found in the index are correct.)

Fix GSF such that during startup indexing, ALL files that register an interest in a file get a chance to index it.
This also fixes 140023: Runtime.php is not indexed in new projects.

132137: NullPointerException at org.netbeans.modules.gsf.GsfDataLoader$1.createFormat

Automated merge with http://hg.netbeans.org/main/

Test fix

Add a configuration object to the structure scanner interface which lets navigator implementations specify whether their contents should be name sorted, filtered, and/or expanded on file open in the navigator. This replaces some hardcoded mimetype logic in GSF. It also lets the new JSON mimetype avoid filtering, and finally we no longer expand ALL nodes in HTML/JSP/RHTML files - only the first level nodes are exposed on open. Should help a bit with performance.

Move JSON to its own mimetype