[clang] [Clang][test] Add bytecode interpreter tests for floating comparison functions (PR #104703)

[Timm Baeder via cfe-commits]

tbaederr wrote:

> @tbaederr I've basically just copied over the tests from the other file. There's no other quad-precision (`__float128`) tests so they were removed. 

Unfortunately that problem also exists with regular floating-point values (provided they heap-allocate memory).

> Do these need to be rewritten in a C++ style? (i.e., replace the macros with template functions)

No, I think this is fine.


The changes work for me locally.

https://github.com/llvm/llvm-project/pull/104703