Why isn't the charge controller using the panel to feed loads when the battery is charged?

There's a very good chance you are getting power from your PV but the controller isn't explicitly telling you so. If in the event your load is larger than your array can supply then of course your batteries will be contributing to power the load. If your batteries are still in absorb your array will be powering the absorb cycle and powering your load with a minimal amount of current into the battery showing. As bill mentioned, If you had a separate monitoring device you could see this. My Tri-metric will show ,for example, a positive charging current of say 21 amps although my Midnite Solar controller shows 40 amps incoming. The difference in current is what is going to my loads at that time. All charge controllers are different in their level of info available too so having the separate monitoring device really helps to know what's going on.

unexpected said:

I'm still watching it like a hawk because I'm programming it.

Good to know you're paying attention, with lithium batteries  it's definitely required.

Most charge controllers are designed for lead acid, setting up the parameters in custom settings is really an adaptation, not really a dedicated lithium charging profile. Each controller will have its individual idiosyncrasies which need to be figured out and as such what works with one won't nesesarally work with another.

As the controller output and load terminals connected in parallel to the battery, once the target voltage is reached charging should terminate within a short duration then transition to float. Being that Lithium batteries have a low self discharge the voltage will remain above the float setting until a load is applied, once the voltage drops and reaches slightly below the float value the controller will begin to pass current to maintain the setpoint. Depending on how often the controller makes its sweeps there may be a slight delay where the battery supports the load, once charging again to maintain the float voltage current will pass to the loads because everything is in parallel.

Minor fluctuations in voltage around the float value shouldn't be considered micro cycles but rather the controllers limited ability to precisely clamp the voltage 100%, what you don't want is large swings where the voltage drops significantly before charging resumes.
More sophisticated controllers will have a recharge setpoint which prompts a fresh cycle once the voltage drops below a certain threshold, usually set at a significantly lower voltage than float. This feature is particularly useful with lead acid which don't tolerate partial states of charge, with lithium that is generally not a major concern, especially in a ballanced system where the array is capable of supporting loads as well as maintain float.

Regarding the 0.5A load not being drawn from the array, how are you determining that? Often controllers won't detect such small values even though they are passing current, try with a larger load. Having a DC clamp on ammeter is almost an essential tool for solar systems, allowing dynamic readings to be made. One additional point the PV voltage input is at a higher value than the battery with MPPT so the 0.5A load at battery voltage, will be less on the PV input, for example if the input voltage is 32V and battery 16V, the 0.5A battery side load would be 0.25A on the PV side as current is inversely proportional to voltage.