[mlir][bufferize] Fix typo in EmptyTensorElimination

[springerm]

Summary

The structure of the code has changed a while ago and the code was not updated properly.

There is no test case for this because we do currently not have an op
that could trigger this bug.

mlir/lib/Dialect/Bufferization/Transforms/EmptyTensorElimination.cpp

for (OpOperand &operand : op->getOpOperands()) {
       SetVector<Value> maybeEmptyTensor = state.findValueInReverseUseDefChain(
           operand.get(), /*condition=*/[&](Value val) { return false; },
           /*followEquivalentOnly=*/true);
 
       // Replace only if the reverse use-def chain ends at exactly one
       // tensor::EmptyOp.
       if (maybeEmptyTensor.size() != 1 ||
           !maybeEmptyTensor.front().getDefiningOp<tensor::EmptyOp>())
-        return WalkResult::skip();
+        continue;
       Value emptyTensor = maybeEmptyTensor.front();
 
       // Replace only if the types match.
       // TODO: This could be extended to support IR such as:
       // %0 = tensor.empty() : tensor<128xf32>
       // %1 = "some_op"(%0) : (tensor<128xf32>) -> (tensor<128xf32>)
       // %2 = tensor.expand_shape %1 ...
       // %3 = tensor.insert_slice %2 into ...
       if (emptyTensor.getType() != operand.get().getType())
-        return WalkResult::skip();
+        continue;
 
       // Find a suitable insertion point.
       Operation *insertionPoint =
           findValidInsertionPoint(emptyTensor.getDefiningOp(), neededValues);
       if (!insertionPoint)
         continue;
 
       // Create a replacement for the tensor::EmptyOp.