Possible memory access error in XC routines

[davidbowler]

When running with non-self-consistent GGA, using the Intel compiler (2020 U4 with OpenMPI 4.1.3, FFTW 3.3.10, MKL 2020 U4) the attached example fails with a SEGFAULT in the force routine (specifically when get_dxc_potential is called in get_nonSC_correction_force but this does not happen with GCC (12.2.0 with OpenMPI 4.1.5 both installed through Homebrew on a Mac).

I'm not completely sure, but it might well be due to the excess use of assumed-size arrays in the XC modules. It would be sensible to tidy these up.

phenol_crash.zip