Assess status of extension type validation checks

[wysiwys]

[wysiwys]

There are some places in the codebase where extension types should likely be checked against the allowed list of types from the RFC, but where these checks are not yet conducted.

• Creating a commit with MlsGroup::update_group_context_extensions()
• Applying a commit created with MlsGroup::update_group_context_extensions()
• Incorrect list is used for validation in MlsGroupCreateConfigBuilder::with_leaf_node_extensions() (#1831)
• Incorrect list is used in group::public_group::TempBuilderPG1::with_leaf_node_extensions() (#1831)
• Extension types are not validated in KeyPackageBuilder::leaf_node_extensions() (#1831)
• Extension types are not validated in KeyPackageBuilder::key_package_extensions()
• Extension types are not checked in LeafNodeParametersBuilder::with_extensions() (#1831)

There may also be some other places where extension types are not checked, or are not conducted consistently.

[franziskuskiefer]

Looks like most of the new builder code didn't implement checks.

There have been proposals before to do this, e.g. openmls/openmls#1138 (comment). I can't recall why we didn't pursue that. But we have to, in general, support any other extension that don't know. So we always need to support unknown extensions in all places, which get all extensions that we don't support natively, and hence are allowed everywhere.

I think having newtypes for constructing messages with extensions, which implements the rfc table makes sense and could be a relatively lightweight change.