Just remember that the continuous load is listed for all the conductors shown in 310.15(B)(16). The values given in that table are indeed continuous loads. However, since we know that OCPD’s have limitations terminals have limitations and so on…
19 Amps on a 12 AWG under the perfect conditions of T310.15(B)(16) would be fine in terms of the conductor itself as again T310.15(B)(16) is based on continuous load values under the tables specific conditions…then the NEC adjusts or corrects accordingly based on ambient, # of conductors, condition of use which change those values given in the table as you know.
Now I would never say 19A on a 20A OCPD would never trip. In theory the trip curves would tend to support that but the ambient rise in heat at the point of termination will affect the OCPD in it’s thermal range. So at some point it will trip even if 19A is placed on a 20 OCPD…but when is the magical question, if the load is continuous at 19A for 3 hours or more…but more than likely a lot more as OCPD’s are robust and do maintain stability well under adverse conditions.
P.S. I have been told by a leading manufacturer of OCPD’s that the above is true, in that the 80% value (125%) is indeed to support the OCPD manufacturers and terminations on those devices. The continuous load effect does result in heat rise to which at some point it overtakes the ambient, and overtakes the long term threshold of the thermal component they will trip at some point…which ironically was the end data that supported FPE in saying they would eventually trip…due to thermal dynamics to the point it would be below thermal ignition points…basically the same theory so to speak.