Your example is trying to demonstrate normal issues that arise during the creation of procs that reference #temp tables. This is easily addressed by insuring #temp table (non)existence prior to creating the proc (alternatively, use DNR). ASE has worked like this for eons ... so no argument that an engineer confirmed this.
My example, and John's issue/question, addresses the creation of nested #temp tables during proc execution (ie, we've already addressed the issues that arise during the creation of the procs).
Once the procs are created (with or withou DNR), it's during execution phase that ASE will create multiple/nested #temp tables ... as per what John's seeing and asking about ... as per my example.