executorch/backends/cadence/reference/operators/quantized_layer_norm.cpp: fix a few format mismatches via static_cast<int>() (#6322)

Summary:
Pull Request resolved: https://github.com/pytorch/executorch/pull/6322

Avoids a few errors like the following:

  executorch/backends/cadence/reference/operators/quantize_per_tensor.cpp:47:55: error: format specifies type 'char' but the argument has type 'ScalarType' [-Werror,-Wformat]

Reviewed By: tarun292, zonglinpeng, mcremon-meta

Differential Revision: D64519629

fbshipit-source-id: baf2f18351ba97fb232dce61b1eb84ebb26216e4