Possible bug in call to tridagMatrix in serial_tri and cycl_laplace

[loeiten]

The call in serial_tri.cxx goes like this

    tridagMatrix(avec, bvec, cvec, bk1d, jy, 
         iz == 0, // Called "int kz" in the function. Used to check if DC.
         iz*2.0*PI/mesh->zlength, // kwave
         global_flags, inner_boundary_flags, outer_boundary_flags,
         &A, &C, &D);

this calls to the invert_laplace.cxx function

void Laplacian::tridagMatrix(dcomplex *avec, dcomplex *bvec, dcomplex *cvec,
                             dcomplex *bk, int jy, int kz, BoutReal kwave,
                             int global_flags, int inner_boundary_flags, int outer_boundary_flags,
                             const Field2D *a, const Field2D *ccoef,
                             const Field2D *d,
                             bool includeguards) {

I think there is a problem here as iz == 0 (iz is the looping index for the mode numbers) will be casted to int kz. As iz == 0 will evaluate to 1 when iz is 0, the result will be that kz = 1 in the invert_laplace.cxx function.
kz will now (as far as I can understand) have a false value, and this value is used to check whether we are dealing with DC or AC components, and it is used if the INVERT_IN_CYLINDER flag is being used.

The same call is being made in cycl_laplace if dst is True.

I thought it would be a good idea to post this as an issue, rather than to just "fix" it, in case I am wrong.