The optimizer will not be forced to use a Product Join because there's another join condition T1.ColB = T2.ColB, thus the LIKE or SUBSTRING might be evaluated during the join.
I would change the LIKE to a SUBSTRING/POSITION, which might enable a direct join on both colA and colB:
T1.ColA=substring(T2.ColA from 1 for position('/' in T2.ColA)-1)
The optimizer will not be forced to use a Product Join because there's another join condition T1.ColB = T2.ColB, thus the LIKE or SUBSTRING might be evaluated during the join.
I would change the LIKE to a SUBSTRING/POSITION, which might enable a direct join on both colA and colB:
T1.ColA=substring(T2.ColA from 1 for position('/' in T2.ColA)-1)