Added a per-stage benchmark measurement harness before moving onto experimental optimisations. To be used to measure future improvements and map changes. - ProcgenStats: time per-stage, number of prefabs placed, placement hash, counters (filter calls, radius scans, texel reads, anchor scans, monument attempts). - ProcgenBenchmark: editor tooling to generate maps for queued seeds and writes a CSV row per-stage. One map per play session, exiting play mode after each generation. - Powershell scripts to drive the editor benchmarks and compare CSV outputs by hash and stage time