That's basically wrong, MGGA can (and does) work fine for many metals. But as always, one should verify it by performing some test calculations. The tutorial you mention was written a long time ago, just after we implemented MGGA and before we had enough experience with TB09 specifically (there are other MGGA functionals like TPSS that are well known to work excellently for metals) and in particular metal/semiconductor interfaces to say with certainty that it worked ok.