As mentioned on this page:
http://www.gammon.com.au/forum/?id=7403
I got the words from this site:
http://wordlist.sourceforge.net/
In particular, it was this file:
http://prdownloads.sourceforge.net/wordlist/scowl-6.tar.gz (2.23 Mb)
They have broken their dictionary lists down into "frequency" files, where the more common, the earlier file number. For example:
american-words.10
american-words.20
british-words.10
british-words.20
To save space and time loading (as well as memory usage) I selected a subset to be included on the downloads page.
It turns out that "violator" is in english-words.50, but I stopped at english-words.40 in the spellchecker file list.
Also "license" is in american-words.20, which also is not in the list.
It won't suggest them if they aren't in the list. Once you load the file, a similar sounding word would suggest "license" - but it needs to be there for the suggestion to be made.
Some of those files are quite small (eg. american-words.20 is only 3 Kb) so maybe they should be in the default list.
I hadn't exhaustively analyzed which files would be required, and in any case it is easy for you to change. If you want to suggest better defaults, without making the spell checker load time be too great, by all means do so.