[Bioc-devel] limma 1.8.18

I've committed limma 1.8.18 to the repository today.  lmFit() is now *much* 
faster for a typical affy analysis, which usually doesn't involve NAs or 
probe-set specific weights. There were some teething problems with special 
cases, 0 or 1 residual df, but I think these are ironed out now. 
Feedback/bug reports welcome.

Gordon