WGSL float % translates into incorrect spirv op code

[Miha-Rozina]

Description
WGSL spec says:

Floating point remainder, where sign of non-zero result matches sign of e1. Component-wise when T is a vector.
Result equal to: e1 - e2 * trunc(e1 / e2)
(OpFRem)

While looking at the shader code in RenderDoc when using Vulkan backend I can see that the modulo operation was translated to:

float _95 = FMod(_92, _94);

Repro steps
Use float % operator in WGSL shader where the first argument is a negative value.

Expected vs observed behavior
The result of the % operation should be a negative value, but it is a positive value, because OpFMod is used instead of OpFRem.

Platform
Using wgpu 0.12.0 on Linux (Arch linux if it matters).