[BUG]: Type cast uint to float

[malex81]

Describe the bug

When cast uint variable to float IL code will be generated :

  IL_0020:  conv.r.un
  IL_0021:  conv.r4

Then ILGPU compiller interprit conv.r.un as cast to float64 (double). And if host (OpenCL) not support double then it's lead to Exception and not work.

Environment

• ILGPU version: 1.5.1
• .NET version: .NET 8
• Operating system: Windows 10
• Hardware Intel(R) UHD Graphics

Steps to reproduce

uint a = ... // any operation that return uint
...
float b = a * 6.18; // any operation that initiates cast uint to float
...

Expected behavior

I think we should consider these two lines as one atomic command

Additional context

No response