tldr: Couple changes needed for Android. Performance isn't spectacular but good enough especially if you don't use a lot of elements.
-----
Overall Emu UI is hands down my favourite UI framework out there for Gamemaker. It looks good. The learning curve isn't bad. Adding new element types is easy. There are a few hardcoded variables you need to change around sizing and padding for a small canvas size like mine (180 x 320) but totally doable. There's also a few functions you need to adjust if you are scaling your game size. If it came with a few extra element types such as a drop-down selection it'd be perfect and as is it is still pretty darn good—especially compared to it's competition on the Gamemaker marketplace, blows everything there out of the water.
You do need to make a couple changes for Android. For example, you need to swap out window_get_mouse_x/y() in the getMouseHover method and a few other places for mouse_x/y or another alternative as the function doesn't work with touch input. Clicking on textboxes also doesn't prompt a keyboard to appear.
Performance is good enough imo. I'm testing it on a Samsung Note 9 and my game, with 0 optimization, gets around 60~ fps with menus/text only going down to 58~ when I draw my sprite-stacked tank.
Performance does steadily drop as you add more elements. I'm not too worried though, with a menu-based game like mine fps doesn't matter too much and I'm probably going to default the game settings to a battery-friendly 30 fps mode anyway on Android. Either way my code has plenty of room for optimization down the line.
For example this EmuDialog: https://imgur.com/rC8y6LG contains 1 EmuButton and 18 EmuText elements (the sprites are being rendered by scribble as part of the EmuTexts' values and are not the fps drop cause) and costs my Note 9 a consistent 5 fps to display. My high-end gaming pc has no issue getting and keeping 144 (fps_real of 500 that drops to 300 with that dialog) while my pretty old non-gaming laptop can touch 144 but drops down to 130~ with that dialog open.
Fair bit of room for optimization though, like I can bring that dialog container down to 4 text elements instead of 18 fairly easily but it does show how performance will degrade with additional elements.
Making a drop-down element is next up on my todo list once I think of how to prevent double-clicking on elements below a list without having to disable any elements.