Pre-computed optimization results (opt_result.json per scenario).

Loaded by the example scripts through the NOADS_RESULTS_DIR environment
variable (see noads.application.examples); docs/conf.py points sphinx-gallery
here so the documentation build never re-runs the optimizations. To recompute
a scenario, delete its directory and run precompute_missing.py from this
directory.

Provenance:
- SSP2-26-* scenarios: computed for the paper (original single_policy results,
  relocated here 2026-07-03).
- SSP1-19-Fossil-*, SSP5-45-Fossil-*, robust-SSP2-midTech,
  robust-SSP2-lowdemand-LowDemand-midTech: computed 2026-07-03 with
  precompute_missing.py on Windows 10, Python 3.11.9,
  jax/jaxlib 0.4.38, diffrax 0.7.1, equinox 0.13.4, lineax 0.1.0,
  optimistix 0.0.11, gemseo 6.3.2, gemseo-jax 2.1.1,
  NLOPT_SLSQP (see noads.application.examples for solver settings).
