Improve input parsing backwards-compatibility

[markleader]

Summary

Improve reac input compatibility with legacy CEA2 behavior, fix #25 so that user-provided reactant enthalpy overrides so they are honored at runtime.

Changes

• source/input.f90

  • Hardened read_input/read_problem flow to avoid EOF/trailing-line parsing crash.
  • Switched read_problem to subroutine-style output and adjusted read_input handling accordingly.
  • Replaced overlapping problems = problems(1:n-1) shrink with allocate/copy/move_alloc.
  • Expanded reac parsing compatibility:
    • case-insensitive token handling for reactant keywords and amount tokens,
    • den... density alias support in addition to rho...,
    • default density units to g/cc when omitted,
    • molecular-weight token support (wt/mol, wt/mole, molwt, mwt, mw) with unit conversion/validation,
    • exploded formula parsing with implicit coefficient 1.0,
    • allow omitted amounts for name reactants,
    • reject mixed mole/weight amount bases in one reac dataset.
  • Fixed formula-token precedence so element tokens (e.g., H, O) are not misclassified as enthalpy tokens.

• source/main.f90

  • Added apply_reactant_thermo_overrides and applied it after reactant mixture creation in all relevant execution paths.
  • Ensures user-provided reactant enthalpy input (h,...) overrides thermo-fit reference values, including for database species.
  • Fixed per-reactant temperature indexing bug in compute_reactant_enthalpy (prob%reactants(i) instead of always index 1).
  • Added defensive defaulting in named-reactant weight assembly when amounts are missing (equal basis with warning).

• source/input_test.pf

  • Added parser regressions for:
    • custom reactant formula + enthalpy + molecular weight,
    • molecular-weight aliases,
    • missing named-reactant amount,
    • density alias/default density units,
    • implicit formula coefficients,
    • case-insensitive reac keywords/tokens.

Testing

• ctest -R cea_core_test -V
• python test_main.py

Compatibility / Numerical behavior

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

Expected numerical/output changes are limited to inputs that explicitly rely on new/changed behavior:

• cases with user-specified reactant enthalpy overrides now honor that input (instead of being overwritten by thermo fits),
• cases with previously invalid/ambiguous reac forms (e.g., missing named-reactant amounts, den alias, molecular-weight aliases, implicit formula coefficients, case variants) now parse/behave in a CEA2-compatible way.