[llvm] [Transforms] Create more optimizing functions to fold inverse trig pairs (PR #77799)

[Yingwei Zheng via llvm-commits]

================
@@ -0,0 +1,54 @@
+; RUN: opt < %s -instcombine -S | FileCheck %s
+
+define float @mytan(float %x) {
+  %call = call fast float @atanf(float %x)
----------------
dtcxzyw wrote:

Same problem as https://github.com/llvm/llvm-project/pull/76641#discussion_r1448163247.

https://github.com/llvm/llvm-project/pull/77799