It is beginning to look like Miss Manners 2009 might (just might, not will) be a good benchmark for checking rules. The other will be the WaltzDB-200. I will be depending on vendors to double check all of my rules for them on that one since the database is generated from WaltzDB.java file that puts it into the "expected" format for the rules. (Thanks to Dr. Forgy for re-writing the WaltzyDB.c file into Java for OPSJ but it can be extended to any rulebase needed.)
Georrrey De Smet has made some changes to the original (hastily done) data file and has blogged about it http://blog.athico.com/2009/05/miss-manners-2009-yet-another-drools.html in case you want to read his blog as well.