WIP: Fix/smooth truncation

[markleader]

Summary

Refines smooth-truncation handling in equilibrium/derivative paths so reported thermodynamic quantities and sensitivities are consistent with post-processed nj values, especially for entropy and pressure derivatives.

Changes

• Updated smooth-truncation linearization in source/equilibrium.f90 (including nj_linear, entropy residual/state terms, and threshold handling per-species including ions).
• Aligned post-processing entropy calculation with reported gas moles/log-moles (nj/ln_nj) and kept condensed-species entropy handling explicit.
• Removed a redundant Jacobian post-assembly adjustment block in EqDerivatives_assemble_J.
• Updated derivative entropy terms to use ln_nj-based expressions in smooth post-processing paths.

Testing

• Not run (prepared PR description from branch diff only; no local test execution in this pass).

• Expanded tests in source/equilibrium_test.pf:

  • switched derivative finite-difference comparison focus to H2O where appropriate,
  • added assertions distinguishing smooth vs hard-truncation derivative behavior,
  • added a new test verifying smooth post-processed entropy consistency and dS/dP finite-difference agreement.

Compatibility / Numerical behavior

• No expected changes to numerical results
• Expected changes (explain and provide validation)
  Expected changes are localized to smooth-truncation/post-processing behavior (entropy and related derivatives near truncation thresholds). Validation is covered by updated/added finite-difference tests in source/equilibrium_test.pf.