Concurrent Chunk Management Engine - Forge

C2ME-Forge 0.2.0-forge.4 (1.18.2)

THE WIDTH RELEASE — r1000 server pregeneration: 96s -> 72s out of the box (-25%),
measured twice with identical results on the same seed.

PERFORMANCE
- Recalibrated default worldgen worker count. The old formula badly under-provisioned
  multi-core machines (a 16-thread server got only 8 workers; clients got even fewer).
  New sizing: ~0.85 workers per hardware thread, bounded by available heap
  (~0.6 GB per worker after a base reserve). Overridable as before via
  globalExecutorParallelism in c2me.toml — if you tuned that by hand, your value wins.
- The adaptive worldgen feed controller and RAM advisor continue to protect small
  heaps and diagnose bottlenecks.

Validated with the same discipline as the 1.20.1 branch: identical formula previously
measured there as 118s -> 100s stock, confirmed at n=3 with interleaved runs.

COMPATIBILITY
- The aquifer optimization's method replacement now uses a lower mixin priority so injections from
  other mods into Aquifer.NoiseBasedAquifer can apply on top instead of failing (the 1.20.1 forge.7
  FDBosses lesson applied preventively).