tilize_op float32 access (#15115)

### Ticket
[Link to Github
Issue](#14885 (comment))

### Problem description
I missed one entry point for tilize when adding float32 support earlier.

### What's changed
- Just added the one entry point in tilize_op.cpp, wanted to push this
out real quick since it's a quick fix and there seems to be a need for
it!

### Checklist
- [ ] Post commit CI passes:
https://github.com/tenstorrent/tt-metal/actions/runs/11862518244

ttnn/cpp/ttnn/operations/data_movement/tilize/device/tilize_op.cpp

@@ -18,7 +18,7 @@ void Tilize::validate(const std::vector<Tensor>& input_tensors) const {
     auto width = input_tensor_a.get_legacy_shape()[-1];
     uint32_t stick_s = width;
     uint32_t num_sticks = input_tensor_a.volume() / width;
-    TT_FATAL(input_tensor_a.get_dtype() == DataType::BFLOAT16, "Error");
+    TT_FATAL(input_tensor_a.get_dtype() == DataType::BFLOAT16 or input_tensor_a.get_dtype() == DataType::FLOAT32, "data type must be bfloat16 or float32");
 
     uint32_t stick_size = stick_s * input_tensor_a.element_size();  // Assuming bfloat16 dataformat