Post-execution state: a fifth layer for the A2A governance stack

The four-layer CTEF proposal (#1786) and the identity extension (#1496) are converging on a clean substrate. A question for the spec: where does post-execution state live?

The current stack covers:
- **identity** — who the agent is
- **transport** — how it communicates
- **authority** — what it's authorized to do
- **continuity** — key rotation and state across sessions

None of these layers describe **what actually happened after authority was exercised** — the terminal execution state that a third-party auditor or a downstream agent needs to verify. An agent can have valid identity, verified transport, scoped authority, and continuous state — and still leave no independently verifiable record of what it did.

The primitive that fills this: a post-execution receipt anchored to an external immutable store, linked back to the authority claim via `delegation_ref` (SHA-256 of the upstream authorization). Not a fifth layer in the identity stack — a complementary accountability layer that the identity stack can reference.

We've built this as [Mycelium Trails](https://github.com/giskard09/argentum-core). Raising here because the spec conversations in #1786, #1496, and #1575 are converging on the right substrate — worth specifying how execution receipts should reference identity claims before the 4.x schema hardens.