Energy diffusion flux applied twice for chemistry + surface_tension without viscous

[sbryngelson]

In s_compute_additional_physics_rhs (src/simulation/m_rhs.fpp) each of the three direction blocks has this shape:

if (surface_tension .or. viscous .or. heat_conduction) then
    do i = eqn_idx%mom%beg, eqn_idx%E          ! <-- includes E
        rhs_vf(i) = rhs_vf(i) + (flux_src(i, prev) - flux_src(i, cur))/d
    end do
end if

if (chem_params%diffusion) then
    do i = eqn_idx%species%beg, eqn_idx%species%end
        ...
    end do
    if (.not. viscous) then                     ! <-- adds E again
        rhs_vf(eqn_idx%E) = rhs_vf(eqn_idx%E) + (flux_src(E, prev) - flux_src(E, cur))/d
    end if
end if

The .not. viscous guard on the second energy contribution was correct when viscous was the only other thing that could take the first branch. It no longer is: with chemistry + surface_tension and viscous = F, the first loop already added the energy source flux over mom%beg:E, and the chem block adds it a second time. The energy diffusion flux is applied twice in every direction.

The guard should match the condition on the first block:

if (.not. (viscous .or. surface_tension .or. heat_conduction)) then

(The heat_conduction combination is rejected at input check as of #1906, so the live case today is chemistry + surface_tension without viscous. The surface_tension half predates that PR.)

Found while adding Fourier heat conduction (#1906).