CUDA minor updates

See merge request !43

One index shift is moved from outside the kernel to inside the kernel.
Shouldn't really change anything. Just making the code more readable
and more consistent with the paper.

Reproducible code

See merge request elpa/elpa!42

Reorganize files

See merge request elpa/elpa!41

Threshold for check pd

See merge request elpa/elpa!40

Now "thres_pd" can be set and used in single-precision calculations.

This is to make the check_pd feature more usable in electronic structure
theory software.

Use `#!/usr/bin/env python3`

See merge request elpa/elpa!39

"#!/usr/bin/python3" only works if python3 is found in "/usr/bin".

Fix for building ELPA with PAPI

See merge request elpa/elpa!37

Loh/python3

See merge request elpa/elpa!36

Only minimal changes done so far (mostly print() stuff)

The associaten with Cuda memory as Fortran arrays was sometimes
done wrongly , since not the number of array elements but instead
the number of array elements times the sizeof(datatype) was used.
This was a mix-up between the C allocation and the Fortran reshape.

This could lead to memory corruption