A few questions about the raylib implementation

[aglebionek]

While playing around with some gamedev using oding+raylib, I've noticed that the raylib Texture struct takes 20 bytes, and the Image takes 24.
Here are the Texture and Image structs to serve as examples, and a helper PixelFormat enum definition.

Texture :: struct {                      Image :: struct {
	id:      c.uint,                     data:    rawptr,
	width:   c.int,                      width:   c.int,
	height:  c.int,                      height:  c.int,
	mipmaps: c.int,                      mipmaps: c.int,
	format:  PixelFormat,                format:  PixelFormat,
}                                         }

PixelFormat :: enum c.int {} // There are around 25 values inside

PixelFormat specifies c.int, taking 4 bytes, however, with only 25 values inside, it could be defined using a single byte. I then noticed that all enums (except Gesture) are defined this way.

Looking again at the structs above, if the width and height properties represent the pixel size, they could probably use just 2 bytes each instead of 4.

My questions are:

1. Is there a reason behind these specific sizes?
2. If the sizes could be lowered, not only for the examples above, but for all enums and structs in the raylib code, would doing so actually make any noticeable difference in the performance?