Update bip39 implementation to @metamask/scure-bip39

[adonesky1]

We are updating all instances of bip39 across our packages to use the @metamask/scure-bip39 (our fork of @scure/bip39) - which allows us to pass around SRP's as Uint8Arrays rather than plain strings for security purposes.

Currently this change may also be blocking: MetaMask/metamask-extension#17056

[FrederikBolding]

The internals of this library will still expect the SRP to be passed as a string. We will need to make further changes to eliminate that assumption.

cc @Mrtenz

[Gudahtt]

LGTM!

[Gudahtt]

Could you elaborate on why this might be blocking that extension PR? I'm having trouble finding a connection.

[adonesky1]

Not going to merge this yet @Gudahtt, @FrederikBolding has indicated that there may be some other adaptations required to make sure this is not breaking. I'm meeting with @Mrtenz tomorrow to discuss.

[adonesky1]

Could you elaborate on why this might be blocking that extension PR? I'm having trouble finding a connection.

I'm not sure if this is going to unblock that but it may be related. The error thrown here in the bip32 test-snap e2e test, is throw here in key-tree. This is happening because the changed bip39 implementation being pulled in in the latest eth-keyring-controller changes are causing the seedphrase to be passed as a Uint8Array rather than a plain text string as is expected in validatePathSegment.