This is motivated by SampleId, whose presence or absence determines whether a
fragment shader runs at sample frequency or not.
In HLSL, if SV_SampleIndex is declared but not used, this results in a signature
entry, but no dcl instruction (and a zero used mask in the signature entry).
Whether the shader will actually run at sample frequency is inconsistent. NVidia
does, AMD does not, and WARP does for d3d12 but not for d3d11.
Previously vkd3d-shader relied on the dcl instruction, thereby aligning with
AMD. This was changed by 66cb2815f0. This commit
restores the previous behaviour.
This makes it more similar to the MSL and GLSL generators. It also looks
like a cleaner design, the backend is supposed to get access to the vsir
program after it has gone through the pipeline.
"char" is (potentially) signed, so casting it to uint32_t will
sign-extend it. Because we use |= to assign it to "word", and don't
otherwise mask out the higher bits either, we effectively set subsequent
bytes in the same word to 0xff for input bytes > 0x7f. That potentially
includes the \0 terminator. For example, "é" (U+00e9) is "\xc3\xa9"
when encoded as UTF-8, and would get us 0xffffffc3 instead of
0x0000a9c3.
