Fix/reflected shock w5

[markleader]

Summary

Fixes #55 with a focus on reflected-shock handling, shock transport output, and graceful behavior around singular/non-converged equilibrium solves.

This branch:

• fixes the reflected frozen u5+v2 discrepancy
• restores shock transport-property computation/output when output transport is requested
• matches CEA2-style shock branch sequencing/early-exit behavior more closely in the CLI
• prevents shock singular/non-converged paths from emitting NaN/stale state
• improves reflected-frozen thermo parity for the reproduced cea_eq case
• adds focused shock regression coverage for transport and singular-recovery paths

Changes

• source/shock.f90

  • fixed reflected frozen shock bookkeeping so the incident rho12 and reflected density ratio are not conflated
  • corrected reflected frozen u5+v2 handling
  • added explicit shock-state validity/failure helpers so failed states are zeroed consistently and stop after the last valid point
  • added shock-specific transport updates, including frozen-shock transport handling
  • finalized retained incident-equilibrium states more defensively after singular/non-converged equilibrium solves
  • added a reflected-frozen finalization path so state-5 thermo is not overwritten by the generic post-processing path

• source/equilibrium.f90

  • added cached transport outputs and cached transport basis state to EqSolution
  • restored cached transport/full-state data on singular or max-iteration exits when available
  • added cp_eq_transport as the equilibrium-transport Cp reported separately from thermo cp_eq
  • avoided rebuilding the transport basis after solves that encountered singular matrices when a cached basis is already available
  • added frozen-shock transport support so frozen shock states compute viscosity/conductivity/Prandtl without equilibrium-reaction transport terms

• source/main.f90

  • updated shock CLI execution to follow CEA2-like branch ordering and early-exit rules for mixed equilibrium/frozen reflected shock cases
  • tracked explicit shock branch codes instead of inferring output sections from permutation index
  • updated shock output so only actually-reached branches are printed
  • added shock transport output blocks in the CLI when requested

• source/shock_test.pf

  • added transport-focused shock coverage
  • added singular-recovery/reflected-frozen regression coverage for the reproduced methane/nitrogen shock case
  • updated failure-path assertions to match the newer CEA2-style recovery behavior

• source/equilibrium_test.pf, source/rocket_test.pf, source/input_test.pf

  • test-only cleanup/debug-output adjustments

Testing

• ctest -R cea_core_test -V

Compatibility / Numerical behavior

• No expected changes to numerical results
• Expected changes (explain and provide validation)

Expected numerical/output changes:

• reflected frozen u5+v2 now matches the legacy CEA2 result for the reproduced frozen case
• shock outputs now include transport sections when requested
• shock singular/non-converged paths stop cleanly instead of printing NaN/stale data
• reflected-frozen thermo for the reproduced cea_eq case is closer to CEA2, especially H/U/G/S and u5+v2

Validation notes:

• cea_fr.out reflected frozen u5+v2 now matches the legacy reference to output precision
• cea_eq.out branch sequencing/output omission now follows the CEA2 path more closely and reflected-frozen G/S are improved