GFortran Issues with LAPACK II

This is an update on my previous post from May. A number of things changed since: GFortran started adopting a fix that by default prevents optimizations which break code calling BLAS/LAPACK functions from C without hidden length arguments. R has been updated to internally add these hidden length arguments (and also in other cases where LTO type mismatch was detected).