So out of curiosity you are saying that the gate charge would drain 'back' through a pull down to ground?
Essentially yes. The gate on a mosfet can be thought of as a capacitor, holding the voltage that is put on it, with respect to ground. It may discharge eventually, through natural leakage, but the resistor will speed the process up. We are talking picofarads of equivalent capacitance, so it's not a lot of charge but it is enough to potentially hold the device on, given the high gate input resistance (i said impedence earlier, my mistake, been awake all night) there is very little in the way of leakage to pull that small charge back to ground and shut the gate off. It may not even be a problem but it is certainly good practice to not leave the gate input effectively floating after the voltage has been removed. It's similar to leaving unused input pins floating on a logic chip for instance, the logic state is unpredictable and just picks up noise or other static charges in the vacinity. Not to mention the floating input could actually be damaged by the high voltage of static electricity.
I hope that helps explain it, to the best of my abilities anyway. I'm sure there are people who can explain it in much more detail, right down to atomic level but hopefully this will suffice enough to get your circuit working close to what you expect.
I have no experience with the WS2811 to be able to comment on that part but i'd be interested to hear your results. I assume it's PWM so the opto's should couple that right through to the mosfets unless it's some absurd frequency above the maximum limits for the opto's and fets. I doubt it on an LED driver though.