feat: [NET-1445] quantum resistant key exchange using ML-KEM

[hpihkala]

Summary

This PR introduces the ability to use (and to require the use of) ML-KEM for key exchange instead of RSA. The former is considered quantum resistant, while the latter is not. This PR does not change the defaults, but introduces a new config option to enable the use of the new quantum resistant algorithm.

Implementing the new algorithm itself was straight forward, but certain refactoring was necessary to use it. Most of the line changes in this PR are actually related to that refactoring instead of the core feature itself.

Backwards compatibility

The change is backwards compatible with the default settings, ie. the change is not breaking as such. However, the new feature can obviously only be used if both the publisher and subscriber have the new version.

Backwards compatibility with the previous version was tested manually, and the results are as expected - ie. key exchange works with default (non-quantum-secure) settings, but fails if requireQuantumResistantKeyExchange is true, as the other side running the old version doesn't support the new algorithm.

Changes

• Implement encryption and decryption using ML-KEM in EncryptionUtil
• Refactor interfaces in EncryptionUtil to be unified across different algorithms
• Add a new StreamrClient config option under the encryption block: requireQuantumResistantKeyExchange: <boolean> (default: false)
• Update SubscriberKeyExchange and PublisherKeyExchange to read that config option and act accordingly
• Add to GroupKeyRequest and GroupKeyResponse protobuf messages an encryptionType field, the value of which is an AsymmetricEncryptionType enum. For backwards compatibility, the absence of this field is interpreted to imply RSA.
• Refactor away the technical debt of OldGroupKeyRequest and OldGroupKeyResponse vs. NewGroupKeyRequest and NewGroupKeyResponse mess and get rid of GroupKeyRequestTranslator. Now, the "new" group key classes (generated from protobuf) are used everywhere, and the old hand-rolled ones have been deleted.
• Add unit test cases for EncryptionUtil and integration test cases for SubscriberKeyExchange and PublisherKeyExchange
• Add a section to the end-to-end encryption page in the docs, describing the new feature
• Add a --quantum flag to the CLI tool to flip on the new switch

Limitations and future improvements

• Refactoring away the OldStreamMessage / NewStreamMessage / StreamMessageTranslator debt is out of scope, the old vs. new classes issue was only cleaned up in relation to the group key classes, as only they were relevant to the feature.
• In the future, encryption preferences could come from stream metadata by default, instead of explicit local config

Checklist before requesting a review

• Is this a breaking change? If it is, be clear in summary. -> (No)
• Read through code myself one more time.
• Make sure any and all TODO comments left behind are meant to be left in.
• Has reasonable passing test coverage?
• Updated changelog if applicable.
• Updated documentation if applicable.

[linear]

NET-1445 Enable key exchange using ML-KEM

[teogeb]

Looks good! 👍 We should keep the EncryptedGroupKey type in the exports.ts, otherwise all ok.