[mlir][GPU] Fix incorrect API usage in RewritePatterns

[springerm]

Summary

Incorrect API usage was detected by D144552.

Depends On: D144549

[guraypp]

Nice that your change can detect this.

mlir/lib/Dialect/GPU/IR/GPUDialect.cpp

LogicalResult matchAndRewrite(WaitOp op,
     if (llvm::none_of(op.getAsyncDependencies(), predicate))
       return failure();
     SmallVector<Value> validOperands;
     for (Value operand : op->getOperands()) {
       if (predicate(operand))
         continue;
       validOperands.push_back(operand);
     }
-    op->setOperands(validOperands);
+    rewriter.updateRootInPlace(op, [&]() { op->setOperands(validOperands); });
     return success();
   }
 };
 
 /// Simplify trivial gpu.wait ops for the following patterns.
 /// 1. %t = gpu.wait async ... ops, where %t has no uses (regardless of async
 /// dependencies).
 /// 2. %t1 = gpu.wait async [%t0], in this case, we can replace uses of %t1 with