We should have an ebuild for OpenSSL 3.0 during the alpha and beta stages to ensure we're ready for release and to track any issues from the fallout. Any such ebuild should be masked because of upstream's clear statement that it is not safe for practical/real life/security-dependnt uses.
After discussion, we're going to wait until a beta is out, because feedback is going to be more valuable for maintainers at that point.