Allow apidoc to skip MDN database.json entries that are bogus.

utils/apidoc/mdn/README.txt

 ***** Current status
 
 Currently it runs all the way through, but the database.json has all
 members[] lists empty.  Most entries are skipped for "Suspect title";
 some have ".pageText not found".
 
 Currently only works on Linux; OS X (or other) will need minor path changes.
 
 You will need a reasonably modern node.js installed.
 0.5.9 is too old; 0.8.8 is not too old.
 
 I needed to add my own "DumpRenderTree_resources/missingImage.gif",
 for some reason.
 
+For the reasons above, we're currently just using the checked-in
+database.json from Feb 2012, but it has some bogus entries.  In
+particular, the one for UnknownElement would inject irrelevant German
+text into our docs.  So a hack in apidoc.dart (_mdnTypeNamesToSkip)
+works around this.
+
 ***** Overview
 
 Here's a rough walkthrough of how this works. The ultimate output file is
 database.filtered.json.
 
 full_run.sh executes all of the scripts in the correct order.
 
 search.js
 - read data/domTypes.json
 - for each dom type:
@@ skipping 36 matching lines @@
 - run postProcess.dart
   - go through the results for each type looking for the best match
   - write output/database.html
   - write output/examples.html
   - write output/obsolete.html
   - write output/database.filtered.json which is the best matches
 
 ***** Process for updating database.json using these scripts.
 
 TODO(eub) when I get the scripts to work all the way through.