Parse WHOIS data more cleanly

The current definition of `__convert_whois_record` assumes certain keys, which are present in the base `WhoisEntry` class, but not in the entries for some other TLDs (for instance, .uk)

I've started to make the code a bit more robust, but it could use a lot more work. 

http://code.google.com/p/pywhois/source/browse/whois/parser.py#37

[From: CybOXProject/Tools#19]
