[mlir][tosa] Cleanup of tosa.rescale lowering to linalg

Comment was poorly written. Changed to bail on contradictory information in
the double round.

Reviewed By: NatashaKnk

Differential Revision: https://reviews.llvm.org/D102651

mlir/lib/Conversion/TosaToLinalg/TosaToLinalg.cpp

@@ -1259,6 +1259,11 @@ public:
     auto outputTy = op.output().getType().cast<ShapedType>();
     unsigned rank = inputTy.getRank();
 
+    // This is an illegal configuration. terminate and log an error
+    if (op.double_round() && !op.scale32())
+      return rewriter.notifyMatchFailure(
+          op, "tosa.rescale requires scale32 for double_round to be true");
+
     if (!outputTy.hasStaticShape())
       return rewriter.notifyMatchFailure(
           op, "tosa to linalg conversion expects statically shaped tensors");

mlir/lib/Conversion/TosaToStandard/TosaToStandard.cpp

@@ -103,7 +103,8 @@ public:
       Value shiftThirty64 = rewriter.create<mlir::SignExtendIOp>(
           loc, rewriter.getI64Type(), shiftThirty32);
 
-      // Round value needs to with be added or sbustracted depending on
+      // Round value needs to with be added or subtracted depending on the sign
+      // of the input value.
       Value roundAdd64 =
           rewriter.create<mlir::AddIOp>(loc, round64, shiftThirty64);
       Value roundSub64 =