[clang] [CIR] Fix lit tests after is.fpclass pretty printing. NFC (PR #209026)

[Erich Keane via cfe-commits]

================
@@ -30,22 +30,22 @@ void test_fpclassify_nan(){
 // CIR: cir.select if %[[IS_NORMAL]] then %[[NORMAL_VAL]] else %[[SUBNORMAL_VAL]] : (!cir.bool, !s32i, !s32i) -> !s32i
 
 // LLVM: %[[VAL:.*]] = load float, ptr
-// LLVM-NEXT: %[[IS_ZERO:.*]] = call i1 @llvm.is.fpclass.f32(float %[[VAL]], i32 96)
+// LLVM-NEXT: %[[IS_ZERO:.*]] = call i1 @llvm.is.fpclass.f32(float %[[VAL]], /* (zero) */ i32 96)
----------------
erichkeane wrote:

ohno, i hate this :D  

I've never actually seen an LLVM printing do something like this.  Thanks for the additional horror added to my life.

https://github.com/llvm/llvm-project/pull/209026