Canonicalize some aspects of the folly/experimental/hazptr test