Oh interesting. So does that mean that

principle_types:1

is inherently worse for operformance tha inlining the wildcards directly on each

has_*

relation? It's sounding like I should eat the maintenance cost and write automation around that, rather than try to use this trick to reduce boilerplate.