We should await confirmation in CDRIVER-4088 that this is indeed the case before making any documentation changes.

Additionally, it does not appear as if libmongocrypt actually requires the BSON binary type to have sub-type 0. We should confirm if that has always been the case and remove any mention of sub-type 0 if so.

Quoting kevin.albertson:

• There was never a subtype requirement.
• Users of libmongoc 1.18.0 can now pass a base64 string as an alternative for binary arguments. In 1.18.0 the only change is that the previous local.key option (which was only acceptable as a BSON binary of any subtype, can now also be configured as a base64 string)