First, lets re-look at the words in 430.62:
430.62 Rating or Setting — Motor Load.
(A) Specific Load. A feeder supplying a specific fixed motor load(s) and consisting of conductor sizes based on 430.24 shall be provided with a protective device having a rating or setting not greater than the largest rating or setting of the branch circuit short-circuit and ground-fault protective device for any motor supplied by the feeder [based on the maximum permitted value for the specific type of a protective device in accordance with 430.52, or 440.22(A) for hermetic refrigerant motor compressors], plus the sum of the full-load currents of the other motors of the group.
The rule is a SHALL BE. So, it is "required". WHAT is required? ? ? The OCPD rating or setting NOT BE GREATER THAN. The rule does NOT say BE EQUAL TO, only "not be greater than". With this in mind, let's revisit your example from earlier:
I believe here is the issue. 430.62 speaks of feeder for fixed motor load(s). This states that a feeder should be based on the largest overcurrent protective device for a motor plus the sum of other motors. We have only one motor so a 3 phase 20 hp motor @ 208 volts is 59.4 amps. Table 430.52 tells us to take 250% of the load for a polyphase inverse time breaker-- 148.5 amps
This means the feeder technically should be 148.5 amps instead of a conductor size for the branch circuit at 125% 74.25 amps
For that 3 phase 20 hp motor @ 208 V, is the OCPD "next size" of 80 Amps "not greater than" 430.52's 148.5 Amps? ? ? Yes.
Or, the OCPD could be 90, 100, 110, 125, etc. as all are "not greater than". The actual OCPD chosen is a design choice. All that the language of 430.62 does is prevent the OCPD rating or setting from going ABOVE your example's 148.5 Amps.
Your argument that the cost of the largest OCPD is too great isn't a SAFETY issue, is it? Isn't the CMP primarily interested in safety only?