FIX: High CPU usage fixes

[ctruxaw]

When the instance is not a Controller (and hence does not use sendAllArtDMX), sendAllArtDMX runs in a tight loop without sleeps or waits. There is no need for this task to run at all for non-controllers.

[ctruxaw]

Found another improvement:
FIX: High CPU usage waiting for SendArtPollEvent
Tight loop caused high CPU usage, instead simply delay until the poll period has elapsed

With these 2 fixes, my CPU usage went from about 40% to 10% when running as a node (I suspect the improvements, specifically for sendAllArtDMX, are less significant when running as a controller)

[patrick-dmxc]

In case the of the Node-Instance its possible to send DMX if there is a Port Configured as Input.

In case of the Task.Delay(), Task.Delay is always langer as the set period, this is getting worth if the CPU is at its limits.
request a delay for exact the Time you want ends in an overshoot.

[ctruxaw]

I'm only using nodes as outputs, hence this fix worked for me. I've updated the patch to instead sleep the thread in the sendAllArtDMX loop

RE: SendArtPollEvent, I'm not sure I understand the concern here. Your tight loop right now pegs the thread to 100% usage... and the timing here isn't critical, even the spec says "It is a requirement of Art-Net that all controllers broadcast an ArtPoll every 2.5 to 3 seconds." - so perhaps change the 2.7 to 2.5, but the timing here should be close enough to be within spec.

These two fixes should cause your code to use about 1/3 the CPU load it did before