(WLEDMM backport) change DDP/E1.31/Art-Net output driver to AsyncUDP (instead of WiFiUDP) for improved performance

[softhack007]

Thanks to another @troyhacks hack, WLED-MM is using AsyncUDP instead of WiFiUDP to improve network LEDs output performance.

We should look into porting this improvement to WLED.

reference source code starts here
https://github.com/MoonModules/WLED-MM/blob/f2d32c9cd80299214ea7966bb1a1afe7cb14e227/wled00/udp.cpp#L917

plus some advanced Art-Net output options here:

https://github.com/MoonModules/WLED-MM/blob/f2d32c9cd80299214ea7966bb1a1afe7cb14e227/wled00/bus_manager.cpp#L482

[troyhacks]

My P4 work also goes beyond this a bit to make it even faster.

AsyncUDP does a bunch of things every time it runs - I have some deeper level code that "assumes starting values until told otherwise" that speeds up the per-packet sending even more. It basically sets up the destination and then skips a bunch of things for subsequent packets.

There's also a lot of dumb routing stuff that the IDF network stack doesn't do well when you mix Ethernet and WiFi - even if you aren't using WiFi. It'll decide to route to the WiFi interface and then it'll get turned around to the Ethernet interface - which is far from ideal. I spent a lot of time in WLED-MM-P4 messing with the routing tables and preferred interface until it uses the correct/best one all the time, and that also required hacking around AsyncUDP.

This is likely more of an issue for me and the work I do, but it is something to be aware of.

You can see the "FastAsyncUDP" overlay class in udp.cpp in my P4 Experimental branch. It is using AsyncUDP but adds even faster packet blasting.