Obviously, recent versions of automake decided to filter out
an "test_" string which led to some surprises

bug fixed: compiler optimization lead to computation 0*NaN=NaN in dssmv. Solved by explicitly setting vector to zero before call of dssmv, not optimal solution. It would be better to make dssmv aware of the beta=0 case.

Some users asked for a more detailed example, on how to set up
_MPI_ and _BLACS_ before using ELPA

This is present in master as well and probably causes the example code
not being gnerated completely by doxygen, as it is seen in the webpages