feat: enforce inbound DPoP in MCP adapters, bound TokenCache, add require_scopes

[muralx]

Summary

Three improvements to the adapters and core, batched into one release:

1. Inbound DPoP proof-of-possession is now enforced end-to-end in the MCP adapters.
2. TokenCache is bounded with a configurable cap + LRU eviction.
3. VerifiedClaims.require_scopes(...) — a new plural AND-style scope helper.

Plus a docs/demo update for the adapter event-loop lifecycle.

Adapters (authplane-mcp, authplane-fastmcp)

DPoP enforcement

• AuthplaneTokenVerifier.verify_token forwards a DPoPRequestContext (method + reconstructed htu + proof header) to AuthplaneResource.verify, so inbound_dpop=InboundDPoPOptions(required=True) checks the proof on every request. The htu origin is always the operator-configured resource URI — never the inbound Host / X-Forwarded-Proto headers.
• htu reconstruction reads scope["raw_path"] to preserve percent-encoding (e.g. %2F) on the wire under ASGI, falling back to request.url.path when the server omits raw_path.

New public surface in authplane-mcp

• AuthplaneRequestContextMiddleware, get_current_request(), and install_request_context(mcp) (idempotent) — an ASGI middleware that publishes the active request on a ContextVar for the verifier. The MCP SDK's TokenVerifier protocol has no per-request hook, so this bridges the request context to the verifier.
• AuthplaneTokenVerifier caches the in-flight verify task per request (keyed by access token on request.state), so a repeat verify_token within one request reuses it instead of re-entering the inbound DPoP replay store. Cross-request replay protection is unaffected.

Note for required=True operators: call install_request_context(mcp) after constructing FastMCP so the verifier can read the per-request context. If it isn't installed, requests fail closed (401) rather than skipping the check — safe by default.

Core (authplane)

• TokenCache is bounded by a configurable max_entries cap (default 10_000, exposed as TokenCache.DEFAULT_MAX_ENTRIES and a read-only cache.max_entries property) with LRU eviction; plumbed through AuthplaneClient.create(cache_max_entries=...). Token-exchange cache keys are high-cardinality (the subject token is part of the key), so the cap keeps long-lived clients bounded.
• VerifiedClaims.require_scopes(scopes: Iterable[str]) — plural AND-style scope-union helper. Empty input is a no-op; on failure the raised InsufficientScopeError carries the full requested tuple on required_scopes and names every missing scope plus the token's available scopes.

Docs / demos

• Examples now run adapter setup, the async server entry point (run_streamable_http_async / run_async), and aclose() in a single asyncio.run(main()), keeping the client's locks, HTTP pool, and background JWKS/metadata refresh tasks on one event loop.

Testing

• ruff check + ruff format --check, pyright (0 errors)
• root 503 passed · authplane-mcp 51 · authplane-fastmcp 51 · conformance 104 passed / 1 xfailed
• python -m build + twine check passed

[RobertoIskandarani]

LGTM