The synopsis seems clear. I also checked the cmcstl2 implementation, which puts this type in the same namespace as everything else from the ranges library, and I asked Casey to review it to get his take as the author.

The intent of P1252 was to restore the usage of dangling from early revisions of P0896, but with a different definition for dangling itself that is more acceptable to LEWG. dangling was defined in std::ranges in P0896r1, so the mistake in P1252 was the detailed specification placing the definition in std and not the synopsis placing the declaration in std::ranges.