Change cgltf_component_read_float so it handles correctly snorm

[tuket]

I have made a small change to the way cgltf_component_read_float converts snorm to float.

The problem was that the lowest value was not being converted to -1.0f.

For example i8 has range [-128, +127].
If you just divide by 127, the mapped range is [-1.007874f, -1.0f].

The solution is simply, if == -128, then return -1.

In addition to that, I have replaced the division for the equivalent multiplication for performance.

Here are a couple of links about snorm conversion:
OpenGL: https://github.com/nlguillemot/SNormTest
DX: https://learn.microsoft.com/en-us/windows/win32/direct3d10/d3d10-graphics-programming-guide-resources-data-conversion
Vulkan: https://docs.vulkan.org/spec/latest/chapters/fundamentals.html#fundamentals-fixedfpconv

I hope it's all good.
Thanks for making this library, it's awesome!

[tuket]

Btw, in cgltf_component_read_float, it's akward that the "default:" case returns 0 and, after the switch, reads the pointer as an integer.

[jkuhlmann]

@tuket The build fails unfortunately.

[tuket]

Sorry... fixed!