[llvm] [LangRef] Adjust reduce.fmin/reduce.fmax behavior for sNaN (PR #196053)

[Zhige Chen via llvm-commits]

================
@@ -21091,11 +21091,12 @@ matches the element-type of the vector input.
 This instruction has the same comparison and ``nsz`` semantics as the
 '``llvm.maxnum.*``' intrinsic.
 
-If any of the vector elements is a signaling NaN, the intrinsic will
-non-deterministically either:
+The reduction is performed in a non-deterministic order. This is only observable
+if one of the inputs is a signaling NaN.
 
- * Return a :ref:`NaN <floatnan>`.
- * Treat the signaling NaN as a quiet NaN.
+For example, if a reduction is performed over ``<sNaN, 0.0, 1.0>``, then all of
+:ref`NaN <floatnan>`, ``0.0`` and ``1.0`` are possible results, depending on
----------------
nofe1248 wrote:

Colon missing in the cross-reference?

https://github.com/llvm/llvm-project/pull/196053