True enough, but how would you get the inverter output to track the load?
It would not be trivial, but IMHO would be do-able.
You could not simply set the inverter for constant voltage output, because you would also need to manage the array voltage so that the power output of the array would match the power consumption of the load.
This might simply 'magically' work if the array voltage is being bucked in the inverter to produce output. If you are on the part of the power curve where current drops if voltage increases (out of the part of the IV curve where the array looks like constant current) then if the load drops, the array voltage will go up and its power will go down. Sounds reasonably stable to me.
If you are on the constant current part of the power curve, then you need an inverter that can deal with the fact that the system is _unstable_; if the output power drops, then the array voltage will increase leading to more power from the array. You would either need a dump resistor, or be very very smart about quickly increasing output voltage (slightly, still in tolerance) to increase output power faster than array power is increasing, and get back to the target steady point.
Hmm. I think I know which part of the power curve I would use
-Jon