Test/legacy

[markleader]

Summary

Adds an opt-in outp ... long mode for the legacy main interface so CEA can print higher-precision numeric output without changing the default formatting path. This branch also updates the main-interface regression fixtures and comparison logic so the tests can validate long-format output reliably when tables wrap across lines.

Changes

• Add a long output flag in the input parser (source/input.f90).
• Update source/main.f90 to emit high-precision ES23.15E3 formatted values when prob%output%long is enabled across the legacy text interface, including equilibrium, rocket, shock, and detonation output sections.
• Introduce shared long-format helpers for reactant rows, ratio lines, trace messages, schedule lines, and rocket station headers to keep the output logic consistent.
• Preserve the existing legacy formatting when long is not requested.
• Switch the test/main_interface example inputs to request long output and refresh the corresponding reference outputs.
• Make test/main_interface/test_main.py more robust for long-format regression output by:
  • aligning repeated/wrapped data blocks before comparison
  • handling the Pinf/P vs Pinj/P parsed key variant
  • comparing the aligned slices instead of assuming identical raw array lengths

Testing

• All repo tests pass.

Compatibility / Numerical behavior

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