Chemistry species entries of flux_src_n are allocated but never deallocated

[sbryngelson]

m_rhs.fpp allocates the per-species entries of flux_src_n under if (chemistry), but the finalize block never deallocates them.

Allocation — src/simulation/m_rhs.fpp, in s_initialize_rhs_module:

if (chemistry) then
    do l = eqn_idx%species%beg, eqn_idx%species%end
        @:ALLOCATE(flux_src_n(i)%vf(l)%sf(...))
    end do

Finalize — s_finalize_rhs_module deallocates three ranges:

• eqn_idx%mom%beg : eqn_idx%E, under viscous .or. surface_tension .or. heat_conduction
• eqn_idx%E alone, under the chem-diffusion condition
• eqn_idx%adv%beg : eqn_idx%adv%end

eqn_idx%species%beg : eqn_idx%species%end is never freed.

Every chemistry run leaks num_species full 3-D fields sized over idwbuff, plus their GPU device mappings. It is a shutdown-time leak, so it will not show up as growth during a run, but it does matter for anything that initializes and finalizes the RHS module more than once, and it is wrong on principle.

Found while adding Fourier heat conduction (#1906); unrelated to that change and present on master.