Enrich M3 sim sub-specs with 11 discovered frameworks and 360:1 minimum speed

- M3 hub: add L3 invariant (360:1 minimum sim speed), six-horizon time table
- M3a: add Heston stochastic vol, rough volatility (fBM), HMM regime detection,
  DCC-GARCH copula, jump-diffusion; six-horizon mapping
- M3b: add Hegselmann-Krause bounded confidence, complex contagion, bandit-
  replicator hybrid, MFG (HJB+FP), pump-and-dump 3-type ABM; six-horizon mapping
- M3c: add six-horizon time table
- M3d: add Kolokoltsov adversarial (non-linear FP + WENO), DSMFG bilevel
  optimization, cross-chain adversarial arbitrage; six-horizon mapping
- M3e: add kinked lending rate model, DeXposure inter-protocol credit network,
  composable yield optimizer; six-horizon mapping
- M3f: add MFG for validator populations, six-horizon time table
- M3g: add Almgren-Chriss optimal execution, six-horizon mapping
- CLAUDE.md: add subagent productivity note

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

core/CLAUDE.md

@@ -28,6 +28,10 @@ Per-unit commands and gotchas live in that unit's `AGENTS.md`. Toolchains (ponyc
 - **S2:** never reclassify a message's provenance.
 - **S3 / vault:** the COBOL invariant-law vault (Invariant 0, the culpability anchor; 01, "harm" "less";) is immutable at runtime — don't edit it unless explicitly directed and only as such.
 
+## Subagents
+
+When spawning background agents (Haiku for research, etc.), **keep working on the main task while they run**. Don't wait idle — fold in results as they arrive, edit other files, or advance unrelated build steps. Background agents are cheap parallelism; wasting the main context window on waiting defeats the purpose.
+
 ## Docs map
 
 - `README.md` — the project and its intent.