Topology, guards, and a finite stability allowance

The hypotheses have concrete meanings. For a matter step they follow from trace-distance contraction and a certified bound

For a coupling step they follow from a declared Lipschitz bound in $d_Q$ and the functional norm of eq:functional-topology. For a geometric step, accessible matter is unchanged. None of these constants or the guard alignment follows solely from thm:spectators,thm:deferred.