>>3243347>well what the test does tell is that the apps are pretty inaccurate
No, it doesn't.
It tells you that the apps are not a suitable replacement for an incident light meter. Not a big surprise since they're NOT incident light meters.
Incident and reflected light metering are two very different things, which is why there's two different types of light meters. It's 100% expected that they're going to give different readings--and different readings in a non-compensatably-uncertain manner--because while the two are related, they're not the same thing. If you have bright light falling on a goth girl clad head to toe in black velvet, and that same bright light falling on a scene girl clad head to toe in gold lame, you're going to get the same reading from an incident meter and vastly different readings from a reflected light meter.
And even between two different reflected light meters, a stop or two difference is not a sign that one is "inaccurate" compared to the other, just that they're using different metering strategies. E.g., Nikon's fancy matrix metering system vs. a simple spot meter are going to give very different results on some subjects. Does that mean one or both of them is inaccurate? No, just means one is trying to be clever about guessing what exposure you actually want based on what the scene looks like while the other is just giving you basic 18% gray.
"Inaccurate" and "performs better" really aren't terms that can apply to metering anyway, since exposure is fundamentally a creative tool. You can't say one meter "performs better" than another because it meters for your main subject to be brighter when you're shooting something backlit because sometimes you want a silhouette and sometimes you want your main subject lit.
Cellphone meter apps are as accurate as any in-camera meter for metering reflected light from a scene. If they weren't, all of your cell pics would be over/under exposed