Validate H5 residue rates on public tool catalogs

Validate the H5 borrowed-signature residue rate on a public tool catalog such as API-Bank.

Background

The paper's controlled ablation and HTB experiments use synthetic registries, and the authors note that the H5 residue depends on the composition of the tool catalog. Although an API-Bank-based result is discussed elsewhere, the authors explicitly identify validation on a public catalog as an outstanding task in the external-validity discussion.

The unresolved question is whether the observed residue rates and the claimed irreducibility of schema-valid borrowed calls generalize to a public, naturally occurring tool catalog.