CoreML maps tensorflow's pad op's symmetric mode to reflect mode.

See here https://github.com/apple/coremltools/blob/0e292a072452db19d1e64b687a372c0c54704a90/coremltools/converters/mil/frontend/tensorflow/ops.py#L1897

The `pad` symmetric mode is mapped to reflect mode, but according to [tensorflow documentation](https://www.tensorflow.org/api_docs/python/tf/pad) the behavior is different - symmetric mode includes the edge values.
So it should report as an error that `symmetric` mode is not supported by CoreML?

Start at coremltools/converters/mil/frontend/tensorflow/ops.py around line 1897 and read the linked TensorFlow pad documentation to compare symmetric and reflect semantics. Trace the surrounding mode handling and confirm the finished behavior reports symmetric as unsupported rather than silently mapping it to reflect.

Written by the indexing model from the issue text.