feat(benchmarks): publish the complete SolverForge 0.22.0 round

Replace the public snapshot with all 5,085 canonical results from the October 18–19 round. Preserve measured SolverForge 0.22.0, Python 0.6.15, Timefold 2.7.1 and PyVRP 0.14.1 provenance, and align the benchmark documentation with the exact selected run IDs.

Keep every solver outcome and official reference visible under the existing feasibility-first ranking. Import existing publishable warehouse evidence only; no benchmark workloads were executed. The full local test and CI gates pass.

Author: blackopsrepl