fix(P3): correct the Cober-List impact-factor sign in the local rime density

[haakon-e]

compute_local_rime_density divides the Cober and List (1993) Eq. 16 impact factor by the signed Celsius temperature. Every subfreezing temperature therefore produced a negative impact factor, the [1, 12] range clamp pinned it at 1, and the local rime density was a constant 159.5 kg/m3 at all temperatures. The parameterization was effectively inert: rime never densified toward the solid-ice limit approaching the freezing point, and all rime-volume (B_rim) sources used floor-density rime.

The fix follows the Cober and List convention with the sub-zero surface temperature bounded strictly below 0 C, min(T - T_freeze, -1e-3), matching the P3 fortran code (microphy_p3.f90, iTc = 1/min(-0.001, Tc)), and applies the sign so the impact factor is positive for subfreezing temperatures.

Impact: at the collision smoke-test state, BCCOL decreases by ~6% and BRCOL by a factor of ~5.7 (large raindrops now produce near-solid rime instead of floor-density rime). The local rime density at fixed sizes now ranges from ~294 kg/m3 at 240 K to the solid-ice limit approaching the freezing point, instead of a constant 159.5 kg/m3. Downstream, this changes the B_rim sources and hence the predicted rime density rho_rim.

Semantics at and above the freezing point (flagged for review): the temperature bound makes the impact factor large, so the clamp returns the solid-ice density. Collisions are evaluated in melt regimes too, and this choice keeps the function finite and monotone there; alternatives (e.g. reusing the 400 kg/m3 default) are possible if preferred.

Tests: interior Cober-List polynomial values, clamp limits, subfreezing densification monotonicity, a pinned value at 240 K, and the freezing-point edge. The two rime-volume smoke-test references are updated to the corrected values; all other collision outputs are unchanged, as expected since only the B_rim components depend on the local rime density.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 92.92%. Comparing base (cd6581c) to head (5ef64cb).
⚠️ Report is 3 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #764   +/-   ##
=======================================
  Coverage   92.92%   92.92%           
=======================================
  Files          56       56           
  Lines        2896     2896           
=======================================
  Hits         2691     2691           
  Misses        205      205           

Components	Coverage Δ
src	93.71% <100.00%> (ø)
ext	69.47% <ø> (ø)

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.