Add dynamic ZCU and ZPPI shadow foundations

[the-code-learner]

Scope

Implements Protocol Economics P1 from the 19 August 2026 working paper while keeping all monetary effects in shadow mode.

Economics and tokenomics

• dynamic per-work-class ZCU reference snapshots from verified compute-slot evidence;
• deterministic weighted medians, EWMA smoothing and bounded per-epoch reference movement;
• ZPPI price-relative basket for compute, data availability and storage with coverage/reliability gates;
• canonical annual ZPPI factor 1_020_408_163, corresponding to the long-run ~2% annual purchasing-power objective;
• public tokenomics documentation explicitly rejects a fixed +2% supply-growth target and treats supply as endogenous;
• ZPPI is the slow monetary-direction signal; ZCSI remains primarily a compute incentive/capacity-routing signal.

Idle-capital P1

• capital-lot lineage primitives, dormancy histograms and capital-weighted productive-coverage accounting;
• deterministic state carrying-cost and fragmentation-pressure simulation;
• prospective IdleCapitalTracker with amount-conserving lineage and explicit productive hooks;
• adversarial/edge coverage for fragmentation, repeated self-cycling, payment/change outputs, mixed-age merges, replay and malformed transitions;
• finalized transaction/result/witness derivation for native-capital movement;
• cross-shard identity uses exactly CrossShardReceipt.Hash() and CrossShardReceipt.DestinationObject();
• tracker is opt-in inside the finalized EpochCollector preview/apply path;
• runtime passes canonical export receipts only after the source state root is known;
• collector checkpoint v2 persists tracker state while retaining restore support for collector checkpoint v1;
• local, cross-shard and restart/canonical checkpoint tests cover the integration;
• finalized compute productive-evidence layer verifies the referenced consumed job object and canonical settlement receipt, reconstructs exact/strict-majority settlement, isolates owner escrow from provider collateral/refunds, and computes overflow-safe paid / escrow coverage;
• productive evidence remains read-only: no whole-job productive mark is applied because job state and settlement outputs can mix escrow with collateral.

Finalized benchmark reporting

• V2 Lab report core counts only finalized-through-consensus samples;
• separates warm-up from the measured window;
• reports finalized TPS and p50/p95/p99 finality latency;
• records shard/validator/workload/environment metadata and JSON output;
• invalidates reports when safety or liveness failures are observed.

This report core is instrumentation, not a new TPS claim. It still needs to be wired into real V2 Lab finality/QC sampling before publishing benchmark results.

Safety boundary

No live ZPH minting, fee split, reward routing, idle levy or other supply mutation is activated by this PR. Generic transfers and arbitrary contract calls do not receive productive-use credit. Thin or unauthenticated economic evidence fails closed.

The implementation rule remains:

measure first -> simulate second -> activate last

Remaining P1

• preserve compute escrow lineage through job creation/assignment/settlement without folding provider collateral into productive credit, then connect only verified paid escrow to productive-coverage updates;
• expand restart/fault scenarios as the shadow tracker accumulates larger state;
• connect the benchmark report to real V2 Lab finality/QC samples and CI/manual matrices;
• run long-horizon public/adversarial economic simulations before any activation proposal;
• keep production-capacity and 1M TPS statements explicitly non-claims until measured through consensus under documented conditions.