opengl: unwanted swscale with 10bit content

Following the changes introduced by 67bddf90

When creating surfaces for 16bpp on the unified loader, the available option uses the combination of: GL_R16UI and GL_RED.

Although the mentioned changes were targeting openGL ES as well as OpenGL, this combination is in fact invalid for OpenGL ES. Note: It is also rejected on my system using Intel OpenGL drivers through EGL.

https://registry.khronos.org/OpenGL-Refpages/es3.0/html/glTexImage2D.xhtml

Because ES does not provide conversions from Integer to Floating point for 16 bits, this combination is always rejected, and in the end makes the caller to fall back using I420. (hence the swscale from I0AL to I420)

To upload designs, you'll need to enable LFS and have an admin enable hashed storage. More information