Add Nirmata Experimental Standard v0.1

[thabatarj8-stack]

What changed

This draft PR adds the first versioned Nirmata Experimental Standard and a machine-readable Experiment #002 package.

It introduces:

• trajectory, run-manifest, blinding, scoring, and publication-safety specifications;
• Experiment #002 protocol v1.0;
• JSON Schemas for traces, predictions, and run manifests;
• independent content and trajectory evaluator contracts;
• holdout design guidance and pre-run checklist;
• run/freeze manifest templates;
• prediction templates and validation notes.

Why

The existing repository explains the Nirmata research thesis and evidence status, but Experiment #002 still needed reproducible operational contracts. This PR moves the project from narrative methodology toward auditable, machine-readable research artifacts while preserving blinding and failure history.

Safety and privacy

The package uses synthetic identifiers and adds an explicit Minimum Necessary Disclosure policy. It contains no ground truth for the fresh holdout and no real names, credentials, private account identifiers, or unnecessary sensitive operational details.

Validation

A pre-PR consistency review aligned evaluator outputs with the prediction schema and aligned the run-manifest schema with the two independent prediction channels. The branch is based on main and contains only additive research artifacts plus consistency corrections within those new files.

Current status

This remains a pre-freeze draft. It does not claim a confirmatory result and should not be interpreted as validation of evaluator performance or the fresh holdout.