Math functions renaming table for GPU backends to support vectorized evaluation of math functions.

[mcourteaux]

Fixes #8594.
This clears up all the preprocessor prologues or wrapper function prologues.

Update: pow() for WGSL is removed from the preamble now, and its behavior is emulated during codegen in the WGSL_CodeGen backend. There is a potential for simplyfing that again if there are meaningful bounds on expressions inferred. I'd consider that future work. Removing this from the preamble and using Halide's IR to emulate this, enables support for vectorized calls. There is a caveat with this: WGSL says there are no nans. However, their own builtin pow() function returns nan for negative inputs to x and y. Emulating that was tricky, because I think the WGSL simplifier removes those? Not sure why I couldn't simply do make_const(x.type(), std::nanf("")). I had to resort to make_const(x.type(), 1.0f) * Call::make(Float(32), "nan_f32", {}, Call::PureExtern) to get the NaN value in.

[mcourteaux]

I seem to have created a few issues. https://buildbot.halide-lang.org/master/#/builders/98/builds/88 (halide-testbranch-main-llvm20-arm-64-osx-cmake) shows useful errors. Will address them.

[mcourteaux]

Is this Windows issue fixed by now?

JIT session error: could not register eh-frame: __register_frame function not found
Assertion failed: !G && "InFlight alloc neither abandoned nor finalized", file C:\build_bot\worker\llvm-20-x86-64-windows\llvm-project\llvm\lib\ExecutionEngine\JITLink\JITLinkMemoryManager.cpp, line 251

@steven-johnson Could you request a rebuild for those failed build bots?

[abadams]

Working on the windows error in #8615

[mcourteaux]

@alexreinking Can you fix mac-x86-worker-2: Python package numpy is no found. If that one is not found, I'm assuming it's a fresh install and more stuff might be missing...

[mcourteaux]

@abadams can you assess the test failure on arm. There is a hexagon related test that failed. I doubt it is related to my work.
@derek-gerstmann can you assess the test failure on Windows? I don't see any error.

[derek-gerstmann]

@mcourteaux The test failure on Windows was the correctness_math test on the Vulkan backend segfaulting due to a failed compilation for mismatched types for built-in constants for inf_f32, neg_inf_f32, and nan_f32. These are handled as extern calls and they assumed they were always scalars, which caused the pow vector x2 tests to fail. I've pushed fixes to make sure proper vector constants are generated.

[abadams]

I will defer to Derek for approval

[derek-gerstmann]

Do we need so many alias macros or is there cleaner way? Otherwise LGTM.