Issue with dipoles: do concurrent and collapse in omp loop

As in the title a user reported an issue with the use of do concurrent in the dipoles, as well as a possible issue with the use of collapse(3). Both in the same subroutine:

https://gitlab.com/lumen-code/lumen/-/blob/develop/src/dipoles/DIPOLE_kb_sum.F#L98 and https://gitlab.com/lumen-code/lumen/-/blob/develop/src/dipoles/DIPOLE_kb_sum.F#L111

My understanding is that both instructions are fine, and to me it looks like a bug of the compiler.