indexing issue in is_jacobian_used()?

[zingale]

in the templated linear algebra, rhs.H, we use reinterpret_cast to change the array indexing to be 0-based, but the linear algebra routines use is_jacobian_term_used<>(), which checks the 1-based species indices defined in namespace Species, so it appears that we have a issue with the indexing.

I think that we need to offset the indexing in is_jacobian_term_used<>() by 1. Note: this will only affect Strang, since with SDC we always return 1.

This could explain some of the differences seen in #1435

[zingale]

turns out that all of the logic was wrong. See #1502.