Drivers and hardware may provide more optimal paths for untyped buffers
compared to typed.
Also, ByteAddressBuffer may be bound via a root descriptor in Direct3D 12,
greatly simplifying the binding logic.
ByteAddressBuffer also doesn't have the 128 * 2^20 element count limit that
typed and structured buffers have, and doesn't require the structure stride
to be specified at resource creation time in Direct3D 11.
Switch between even and odd 16-byte element sequences along X by simply
flipping a bit rather than going to a different resolution-scaled group of
pixels, by increasing the size of the group within the constraints imposed
by tiling.
Use the _xe suffix instead of the xesl_ prefix for quicker visual
recognition of identifiers, also switch to snake_case for consistency.
Also add the f suffix to float32 literals because the Metal Shading
Language is based on C++.
Ordering the descriptor sets by the change frequency on Vulkan, in increasing order (the opposite of D3D12 root signatures). The EDRAM binding never changes there (always one storage buffer), while the destination buffer binding may become changeable in the future (to split dispatches if exceeding `maxStorageBufferRange`, for example).
