[llvm] [llvm] Improve TLI for Darwin libsystem_m functions (PR #109479)

[Tex Riddell via llvm-commits]

================
@@ -0,0 +1,43 @@
+; RUN: opt -vector-library=Darwin_libsystem_m -passes=inject-tli-mappings,loop-vectorize -S < %s | FileCheck %s
+
+target datalayout = "e-m:o-i64:64-i128:128-n32:64-S128"
----------------
tex3d wrote:

Which would you prefer?  I think the options are:
1. Remove atan2 additions from this PR, I add those to my PR.
2. I remove `isTriviallyVectorizable` atan2 change from my PR and my PR becomes about adding the missing massv llvm.atan2 intrinsic mapping after this PR merges.

I do wonder what's going on with the cases added to `hasOptimizedCodeGen`, since there are no tests failing without these additions, nor are there new tests verifying that these additions make any difference.  If you were to remove the atan2 case, there wouldn't be any tests updated in this PR verifying the change.  I'm afraid I'm not familiar with the impact that the missing cases from `hasOptimizedCodeGen` had.

https://github.com/llvm/llvm-project/pull/109479