Add a bond-type-aware bound. The current bounds mainly use fragment sizes and eligible edges. The assembly-theory implementation also exploits distinct bond types. I derived its applicability here and checked it against 45,014 exact residual states: no inadmissible results, and it tightened the generic scalar bound in 2,229 states (4.95%). This warrants a benchmark; it does not yet demonstrate improvement over the strongest production bounds or faster execution.
Add a bond-type-aware bound. The current bounds mainly use fragment sizes and eligible edges. The assembly-theory implementation also exploits distinct bond types. I derived its applicability here and checked it against 45,014 exact residual states: no inadmissible results, and it tightened the generic scalar bound in 2,229 states (4.95%). This warrants a benchmark; it does not yet demonstrate improvement over the strongest production bounds or faster execution.