feat(adapters/msteams): land Microsoft Teams adapter as experimental (needs live-tenant validation)

Track landing the Microsoft Teams adapter (`adapters/msteams`) at the **experimental** tier, per ADR 0007.

## Status

A high-quality spike exists as draft PR #4: self-contained (zero core-runtime changes, zero new dependencies), stdlib-only Bot Framework JWT/JWKS verification, versioned opaque thread IDs carrying `conversationReference`, reuse of `internal/ratelimit` and the `Observer` hooks, ~78% statement coverage against a mocked Bot Framework server.

The spike **resolved ADR 0007 open question 9** (no `msbotbuilder-go`; hand-rolled stdlib JWT validation works, no new deps). The remaining **9 of 10 open questions require a live Azure Bot registration + Teams tenant** to validate (ack/turn semantics, endorsement behavior on production keys, markdown fidelity, `serviceUrl`/conversation-id stability, proactive posting prerequisites, `aadObjectId` stability, etc.). See the 10-point checklist in PR #4.

## Path to land

1. A human with Azure Bot + Teams tenant access runs the live validation checklist from PR #4.
2. Update ADR 0007 → `Accepted`, recording confirmed findings.
3. Mark the package experimental in `adapters/msteams/doc.go` (honest tier labeling).
4. Undraft PR #4, pass CI + `@codex review` + `@codex security review`, merge.

Blocked on: live-tenant access (human).

---
_Generated with [`mux`](https://github.com/coder/mux) • Model: `anthropic:claude-fable-5` • Thinking: `xhigh`_

Inizia con la checklist di validazione live in 10 punti della PR in bozza #4 e dell'ADR 0007; ciò richiede l'accesso a una registrazione Azure Bot e a un tenant Teams. Registra i risultati confermati nell'ADR 0007, contrassegna adapters/msteams/doc.go come sperimentale, quindi verifica CI e le review richieste prima di fare il merge della PR #4.

Scritto dal modello di indicizzazione a partire dal testo della issue.