More SOC on GPU

Description

Move the operations for SOC to GPU to avoid unnecessary CPU-GPU data transfers.

• adapt bra and ket kernels for complex values
• add new kernel for complex mixing
• adapt projector_bra_force_complex kernel to complex matrices
• add profiling region for mixing part of code
• use single exit point for X(hamiltonian_elec_base_nlocal_start) and X(hamiltonian_elec_base_nlocal_finish) subroutines
• adapt SOC test (linking to scalapack, and AllowCPUonly option are not required anymore)

Checklist

• I have checked that my code follows the Octopus coding standards
• I have added tests for all the new features added in this request.