[SYCL][E2E] Ensuring lowering of llvm.bitreverse is functionally correct

[LU-JOHN]

Ensure that lowering of llvm.bitreverse* intrinsics by llvm-spirv is functionally correct.

[sarnex]

I will leave in-depth review of the test logic to others.

[sarnex]

Do we really need to pass O2 here?

[LU-JOHN]

Optimization is necessary to ensure that functions are recognized as having bitreverse functionality and optimized in LLVM IR to llvm.bitreverse.* intrinsics.

[LU-JOHN]

-O2 would not be necessary if we could use __builtin_elementwise_bitreverse for all cases, but there is an issue with this builtin for 8-bit and 16-bit types.

[LU-JOHN]

There are a few remaining questions/issues with this testcase:

1. This testcase is a draft because it is waiting for the llvm-spirv changes to be pulled in.
2. This testcase cannot use __builtin_elementwise_bitreverse on 8-bit and 16-bit data. This testcase has a workaround to avoid this issue: Unexpected codegen for __builtin_elementwise_bitreverse on unsigned char and short  llvm/llvm-project#84047
3. This testcase also builds without explicitly disabling SPV_KHR_bitinstructions (i.e. the straightforward compile options). This fails to execute because of an unsupported capability. We recently enabled SPV_KHR_bitinstructions in the driver: [SYCL] Enable SPV_KHR_bit_instructions for SYCL code #12754 . Is this safe?

# .---command stderr------------
# | terminate called after throwing an instance of 'sycl::_V1::compile_program_error'
# |   what():  The program was built for 1 devices
# | Build program log for 'Intel(R) FPGA Emulation Device':
# | Compilation started
# | Unsupported SPIR-V module
# | SPIRV module requires unsupported capability 6025
# | Compilation failed
# |  -11 (PI_ERROR_BUILD_PROGRAM_FAILURE)
# `-----------------------------
# error: command failed with exit status: -6

[LU-JOHN]

llvm-spirv changes to lower llvm.bitreverse.* have been pulled in. Thus, test is no longer a draft and ready for review.