it makes a lot of sense IMHO. it abstracts away dependency on exact SQLite version, but preserves a familiar SQL-like mental model. it is stable, mature, well understood and widely deployed/used (god i sound like a shill.. ;)
and has the added benefit that it can be implemented in two different flavours: as a standard JavaScript (ES3) library on top of WebSQL (or SQLite in Gecko), or as a first-class language extension with syntax sugar similar/related to array comprehensions (from JS1.7/Harmony/ES6)..
check out the two code samples, and a list of language extension (all feeling right at home in JS): http://en.wikipedia.org/wiki/Language_Integrated_Query#Langu...
if it wasn't proposed, can someone from WHATWG (like you Brendan ;) politely ask Microsoft to "contribute" any related intellectual property to W3C? how likely do you think would it be for AAPL/GOOG to accept it? (and if not, it could simply be implemented as a thin layer above WebSQL ;)
I personally think there's definitely room for a SQL database API built into the browser, but Mozilla did have a point. It should be something a little bit better specified than that.
Also, it's not singlehanded -- IE refused to support Web SQL as well IIRC, and they aren't going away anytime soon.