Now my understanding is that even though an antenna is Zero reactance is still may not be resonant on certain frequencies. How is this possible?
Let's state some principles to see where the confusion is coming from. From The IEEE Dictionary
"resonance (5)(A) (radio-wave propagation)
The rapid increase or decrease of the (signal) amplitude as the excitation frequency approaches one of the natural frequencies of the system."
For resonance, we must have an inductive reactance and a capacitive reactance that are equal in absolute magnitude such that energy is being exchanged between the two types of reactances. In other words, the two different types of reactances are neutralizing each other such that the total reactance in the system adds up to zero leaving one with a pure resistance. That's what we are doing when we adjust a tuner for a 50 ohm Z0-match at the tuner input. In a low-loss system, when we achieve that Z0-match, we are causing the signal amplitudes to peak at the antenna thus radiating the maximum available power.
An ideal dummy load presents a purely resistive impedance but is not resonant because it does not meet the above definition.
The purely resistive value at resonance can have any value, e.g. 1 ohm, 10 ohms, 50 ohms, 100 ohms, ... There is nothing magic about 50 ohms.
Almost all SWR meters are calibrated for 50 ohms and the SWR reading will be 1:1 only when a value of 50 ohms exists. A 1 ohm resonant circuit will indicate a 50:1 reading on an SWR meter. A 100 ohm resonant circuit will indicate a 2:1 reading on an SWR meter. A 50 ohm SWR meter is worthless for determining resonance except for a 50 ohm value at resonance, i.e an SWR meter cannot be used to detect resonance at any other resonant resistive value except 50 ohms
A grid dip meter can be used to detect resonance at virtually any resonant resistive value.
On a 50 ohm SWR meter, a resonant value of 10 ohms will indicate an SWR of 5:1. With a non-resonant value of 20+j20 ohms, it will indicate an SWR of 3:1, i.e. a non-resonant value of impedance can give a lower SWR reading that a resonant circuit.
Point is: When we adjust our antenna systems for lowest SWR, we may not be adjusting them to resonance.
An ideal system-wide conjugate match exists only in a lossless system which is impossible in the real world.
A system-wide near-conjugate match can exist in a low-loss real world system.
High losses in a real-world system prevent a system-wide conjugate match from being achieved. For instance, if the resonant impedance looking into the transmission line is one ohm, almost half the power from the transmitter will be dissipated in the tuner and a lot more in the coax with the 50:1 SWR.
Hopefully, I have touched on the source of the confusion.