[llvm] [LLVM][IR] Add constant range support for floating-point types (PR #86483)

[NAKAMURA Takumi via llvm-commits]

https://github.com/chapuni commented:

This has introduced a layering violation.

Pruning the problematic method may be a quick fix. Or please revert atm.

https://github.com/llvm/llvm-project/pull/86483