possible bug in returning wavefunction amplitudes #19

_solve_sci_core reshapes the dumped wavefunction against the carryover determinant counts, but SaveMatrixFormWF writes it over the full input subspace. The sizes disagree here, so the code silently substitutes a uniform amplitude array. as a result, any downstream property computed from the amplitudes becomes wrong.

include/sbd/chemistry/tpb/sbdiag.h:753 dumps over adet/bdet, not co_adet/co_bdet:

SaveMatrixFormWF(sbd_data.dump_matrix_form_wf,
                 bit_length, L, adet, bdet, ...);

with the default (carryover_type=1, ratio=0.1) the carryover keeps ~10% per spin, so there's a mismatch.
I think the fix should be to reshape against the full subspace regardless of the carryover_type, which to my understanding is relevant for the next SQD iteration.
If you agree, I'm happy to open a PR for it.