Integer division by zero error in sparse trsv

[MalachiTimothyPhillips]

I have a use-case where a matrix has no entries on a given processor. This ends up doing a division by zero (the local row count) here: https://github.com/kokkos/kokkos-kernels/blob/develop/sparse/impl/KokkosSparse_trsv_impl.hpp#L212.

On some platform/compiler combinations, this ends up becoming a Floating point exception: integer divide by zero.