skip one op for coreml partitioner (#2826)

Summary:
Pull Request resolved: https://github.com/pytorch/executorch/pull/2826

There are some issues with the current two ops in coreml side. While waiting for the fix, skip them for now
```
"aten.index_put.default"
```
test with
```
python3 -m examples.models.llama2.export_llama --coreml --use_kv_cache
```

Reviewed By: mergennachin

Differential Revision: D55680297

fbshipit-source-id: 99225a22b0369583c36fe67709b66b0784a7765d