color attachment byte cost validation panics for native only formats like Rg16Unorm

[ScanMountGoat]

I'm using 16 bit normalized color attachments on desktop with wgpu::Features::TEXTURE_FORMAT_16BIT_NORM. This was working in 0.19.0 but panics in validation code on the latest release.

wgpu/wgpu-core/src/validation.rs

Line 1262 in e26d2d7

let byte_cost = format.target_pixel_byte_cost().unwrap();

It looks like the validation code only includes formats supported on the web.

wgpu/wgpu-types/src/lib.rs

Lines 3629 to 3663 in e26d2d7

	pub fn target_pixel_byte_cost(&self) -> Option<u32> {
	match *self {
	Self::R8Unorm | Self::R8Uint | Self::R8Sint => Some(1),
	Self::Rg8Unorm
	| Self::Rg8Uint
	| Self::Rg8Sint
	| Self::R16Uint
	| Self::R16Sint
	| Self::R16Float => Some(2),
	Self::Rgba8Uint
	| Self::Rgba8Sint
	| Self::Rg16Uint
	| Self::Rg16Sint
	| Self::Rg16Float
	| Self::R32Uint
	| Self::R32Sint
	| Self::R32Float => Some(4),
	Self::Rgba8Unorm
	| Self::Rgba8UnormSrgb
	| Self::Bgra8Unorm
	| Self::Bgra8UnormSrgb
	| Self::Rgba16Uint
	| Self::Rgba16Sint
	| Self::Rgba16Float
	| Self::Rg32Uint
	| Self::Rg32Sint
	| Self::Rg32Float
	| Self::Rgb10a2Uint
	| Self::Rgb10a2Unorm
	| Self::Rg11b10Float => Some(8),
	Self::Rgba32Uint | Self::Rgba32Sint | Self::Rgba32Float => Some(16),
	Self::Rgba8Snorm | Self::Rg8Snorm | Self::R8Snorm => None,
	_ => None,
	}
	}