docs: Add Sigstore integration guide, threat model, identity lifecycle, and migration guides

[bordumb]

Context

The Sigstore/Rekor integration is working (DSSE entries, PEM keys, inclusion proofs). The curve-agnostic architecture is documented in docs/contributing/adding_curves.md. But several user-facing and security-critical docs are missing or scattered.

Four docs needed

1. Sigstore integration guide (docs/guides/sigstore.md)

User-facing guide for the --log sigstore-rekor feature. Should cover:

• What happens when you run auths artifact sign --log sigstore-rekor
• What the DSSE envelope contains and why we use DSSE instead of hashedrekord
• How to verify your entry with rekor-cli get --log-index <N>
• How to search Rekor by public key
• What the transparency section in .auths.json means
• How the CI path (--ci --log sigstore-rekor) differs from the device-key path
• What --allow-unlogged means and when to use it

Currently this information is split between docs/design/transparency-log-port.md (implementation-level) and ADR-007 (decision rationale). Neither is a "how to use it" guide.

2. Sigstore threat model (docs/security/sigstore-threat-model.md)

Security analysis of the Rekor integration boundary. Should cover:

• What happens if Rekor is compromised (split-view attack, how inclusion proofs defend)
• What happens if Rekor is unavailable (signing still works, logging fails, --allow-unlogged)
• Trust dependency on the Linux Foundation's operational security
• GHSA-whqx-f9j3-ch6m countermeasure (verify response matches submission)
• Why we don't use Fulcio and the security implications of skipping it
• The DSSE signature vs attestation signature distinction — which signs what
• Rate limiting and retry behavior (429 handling, single retry, exit code 4)

The existing STRIDE threat model (THREAT-MODEL-001-stride.md) covers six trust boundaries but doesn't include the Rekor boundary.

3. Identity lifecycle guide (docs/guides/identity-lifecycle.md)

End-to-end walkthrough of managing an auths identity. Should cover:

• auths init — what happens under the hood (key generation, KERI inception, keychain storage)
• Key rotation — auths id rotate, pre-rotation, what happens to old commits
• Device pairing — auths pair, QR flow, multi-device signing
• Device revocation — removing a compromised device
• Recovery from device loss — what you can and can't recover
• Attestation expiry and renewal (when TTLs land)

This is currently scattered across CLI --help text, the KERI design doc, and the pairing protocol doc. No single narrative.

4. Migration from GPG/Sigstore (docs/guides/migrating.md)

Practical steps for developers switching from an existing signing setup:

• From GPG: how to keep old GPG-signed commits verifiable while switching new commits to auths, what to do with your GPG key
• From Sigstore/Gitsign: how to switch from gitsign to auths-sign as the signing program, what changes in verification, how Rekor entries differ
• From unsigned commits: the simplest path, just auths init

The blog post "The Three Paths to Signing a Commit" compares the three approaches but doesn't give migration steps.

Scope

All four are Markdown files in docs/. No code changes. Add entries to mkdocs.yml under the appropriate nav sections.