Disable internal fp16 llama unittest

Summary:
Something non-xnnpack is falling on to portable and it is failing to handle it.

```
[op_scalar_tensor.cpp:33] In function operator()(), assert failed (false): Unhandled dtype Half for scalar_tensor.out
```

Created from CodeHub with https://fburl.com/edit-in-codehub

Reviewed By: JacobSzwejbka

Differential Revision: D55332891

fbshipit-source-id: 8dcbd79b15aee1e0f1f5f108a77be8e410dab5ce

Author: digantdesai

backends/xnnpack/test/models/llama2_et_example.py

@@ -16,6 +16,7 @@ class TestLlama2ETExample(unittest.TestCase):
     def test_f32(self):
         self._test()
 
+    @unittest.skip("T183420542: Add proper fp16 support.")
     def test_f16(self):
         self._test(torch.float16)