[IRBuilder] Is strict FP mode reasonable?

I'm working on ⚙ D53157 Teach the IRBuilder about constrained fadd and friends and need community feedback please.

The ticket is adding the beginnings of strict floating point / comstrained FP intrinsics. The current thinking is that it would be better to have a strict mode such that, for example, a call to CreateFAdd() will get you the constrained intrinsic when in strict mode. This means front end changes would be smaller.

Does this seem reasonable for clang?