-also remove a type in variable name

The same should be done for the complex case

The AVX2 kernels might not have been called, depending on how
ELPA has been build

The thread_width was not set correctly

a preprocessor flag was missing

The AVX2 kernels might not have been called, depending on how
ELPA has been build

The thread_width was not set correctly

a preprocessor flag was missing

This closes issue #23