Performance issue: `glGetBufferSubData` calls taking a long time

[bravely-beep]

With Effective installed, 12% of my render thread time is being spent on a libc __GI___ioctl() call, which is called by a glGetBufferSubData in Veil's light rendering code. I don't know much about GPU programming, but a quick google search suggests this could be a resource contention stall.

Spark trace:

This is causing a noticeable FPS hit, so probably worth looking at.

Using Effective version 2.4.8-1.21.1 on Fabric.