fix(channel): normalize low-order peer key errors

[dinakarjs]

What

Normalize X25519 exchange failures in SealedChannel.seal() to SealedChannelError, preserving the backend exception as the cause. The catch covers the exchange operation only. Add three low-order-key regression cases and a changelog entry.

Why

A 32-byte low-order public key passes key parsing but raises a raw backend ValueError during exchange. Callers handling the channel's documented error family cannot handle this consistently with other invalid keys or open failures.

Security impact

The low-order keys already fail closed before ciphertext is created. No encryption format, key acceptance, attestation binding or key-custody guarantee changes. This is error normalization, not a confidentiality bypass fix. The channel path requires two maintainer approvals under CONTRIBUTING.md.

Test plan

• Before the fix: 3 new cases fail with raw ValueError; 7 existing channel tests pass.
• After the fix: full unit suite, Python 3.12: 793 passed, 2 skipped, including all 10 channel tests and existing round-trip/wrong-key/tamper/AAD controls.
• Ruff src/tests, mypy runtime/verify (47 files), Bandit and diff whitespace checks pass.
• Tested local tree 6cee90c4918aa69768d87ad46992c8285c54afe8 matches the published tree. No hardware experiment performed.

DCO sign-off

• I certify that I wrote or have the right to submit this contribution, and I agree to the Developer Certificate of Origin. Commit signed off.

[github-actions]

🟡 Contributor Check: MEDIUM

Check	Result
Profile	MEDIUM
Credential	LOW
Overall	MEDIUM

Automated check by AgenTrust Contributor Check.

[codecov-commenter]

⚠️ Please install the to ensure uploads and comments are reliably processed by Codecov.

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[noah-ing]

Reviewed exact head fee80aa against 305fedc. No actionable defect found. The catch covers only X25519 exchange ValueError, preserves its cause and detail, and normalizes already-rejected keys into the documented SealedChannelError.

Fresh Linux/Python 3.12 environment using the hash-pinned lock: the final full unit/conformance run passed 846 tests, with two real-hardware capture tests skipped. Independently authored checks passed 29 cases, including rejected encodings, exception boundaries, and deterministic base/head ciphertext compatibility. Ten normalization cases fail against the base with the expected raw ValueError.

Ruff, format, mypy, Bandit, dependency consistency and dependency audit pass. Source stayed unchanged; DCO trailer and clean merge simulation confirmed. The PR description accurately limits the security impact to error normalization. This is software/error-contract validation, not hardware-attestation evidence; the required two independent current-head maintainer approvals remain separate.

[rajnisht7]

the changes looks good, low order X25519 keys that parse but fail at exchange() now raise SealedChannelError with the backend ValueError preserved as cause, same as the existing invalid-key path. catch is limited to the exchange call, no ciphertext is produced, and the three regression cases cover it cleanly

[imran-siddique]

Approved fee80aa: the three low-order cases fail with main's sealed.py and pass here, and the change only wraps the X25519 exchange. channel/ is a security path, so this needs a second approval. @carloshvp, could you review it by Monday 19 October?