Re-enable `test_save_load_dqtensors` in `test/integration/test_integration.py`

[jerryzh168]

Currently it's skipped because we can't save local lambda function, that is used in LinearActivationQuantizedTensor:

ao/torchao/quantization/quant_api.py

Line 452 in 28f48fd

input_quant_func = lambda x: to_affine_quantized(x, input_mapping_type, get_per_token_block_size(x), input_target_dtype, eps=input_eps, quant_min=input_quant_min, quant_max=input_quant_max, scale_dtype=torch.float32 if x.dtype == torch.float16 else None)