get_resource_types_by_perm should fully leverage caching

[whatisgalen]

get_resource_types_by_perm can get called many times when assigning permissions to search results. It calls get_nodegroups_by_perm which does leverage caching, but then proceeds to do an ORM query every single time it's called.

def get_resource_types_by_perm(
        self, user: User, perms: str | Iterable[str]
    ) -> list[str]:
        nodegroups = self.get_nodegroups_by_perm(user, perms)
        graphs = (
            Node.objects.values("graph_id")
            .filter(
                Q(nodegroup__in=nodegroups)
                & ~Q(graph_id=settings.SYSTEM_SETTINGS_RESOURCE_MODEL_ID)
                & Q(graph__isresource=True)
            )
            .values_list("graph_id", flat=True)
        )

        return list(str(graph) for graph in graphs)

[whatisgalen]

@philtweir interested if F&T had a different approach when it came to this ORM query in get_resource_types_by_perm