IGR silently drops qv (heat of formation) from the pressure, and nothing prohibits qv with igr #1778

src/simulation/m_igr.fpp:412 declares

real(wp) :: qv_igr  !< discarded; IGR carries no heat of formation

and passes it as the output slot of s_compute_mixture_coefficients at ten sites, discarding the
result each time. Consistent with that, the IGR CFL routine computes pressure without qv
(m_igr.fpp:2591-2592, :2605-2606):

pres_L = (E_L - pi_inf_L - 0.5_wp*rho_L*(vel_L(1)**2._wp + vel_L(2)**2._wp))/gamma_L

where the shared operator is f_pressure(e_int, gamma, pi_inf, qv) = (e_int - pi_inf - qv)/gamma.

Nothing enforces the invariant the comment asserts. There is no rule in
toolchain/mfc/case_validator.py linking igr to qv — the IGR prohibitions are only about
igr_pres_lim, igr_order, non_newtonian, num_igr_iters, igr_iter_solver and down_sample.
So a case may set fluid_pp(i)%qv to a non-zero value together with igr = T, and the IGR pressure
will silently omit the heat of formation. Wrong pressure, no warning.

Current exposure is nil — none of the seven examples/*IGR* cases sets qv — but that is luck, not
a guard.

Two ways to close it, and the choice is a physics one:

1. Enforce the documented invariant: self.prohibit(igr and any(qv != 0), "igr does not support heat of formation (qv)") in case_validator.py, with a PHYSICS_DOCS entry. Minimal, and makes
   the existing comment true.
2. Support it: thread qv_igr into the IGR pressure (and the corresponding energy), at which
   point qv_igr stops being discarded and the comment goes away.

Found while auditing the last un-centralized equation-of-state expressions for #1762 / #1638: this
routine is the only place left in src/ that writes the stiffened-gas pressure and sound speed
inline rather than calling the shared operators.