Except grouping classes from rule that top-level classes should be in COB

[sebastianduesing]

During the 2025-06-02 COB development call, we discussed implementing a policy that was proposed in the past that would except grouping classes (i.e., terms for a union of n other terms) from the rule that COB-aligned ontologies should have all their root terms in COB, iff all classes included in the union are in COB. @nataled raised this as a solution for the issue of PRO's grouping class "protein-containing molecular entity", but this exception would obviously extend to others.

This issue is for discussion of this exception and logistics of its potential implementation (e.g., where and how this principle/exception should be published).