[quant][core][improvement][bug fix] Added channel axis bound checking in fused_moving_avg_obs_fake_quant_*
Summary:
Addresses issue raised in #77896. Previous code did not check for bounds on channel axis in fused_moving_avg_obs_fake_quant_cpu/cuda, resulting in undefined behavior. Changes in this PR adds bound checking.
Test plan:
```
python test/test_quantization.py
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/78148
Approved by: https://github.com/jerryzh168