request GDN to support inference

Tagging @mcore-oncall for visibility.

**Is your feature request related to a problem? Please describe.**
We are using Megatron-Core's experimental `gated_delta_net` attention variant in a GPT decoder model for training, but inference is currently blocked by an explicit `NotImplementedError` in [`megatron/core/ssm/gated_delta_net.py`](https://github.com/NVIDIA/Megatron-LM/blob/main/megatron/core/ssm/gated_delta_net.py).

The current implementation rejects inference unconditionally:

```python
if inference_context is not None:
assert (
inference_context.is_static_batching()
), "GDN does not currently support dynamic inference batching."
assert not self.config.sequence_parallel
# TODO: support inference
raise NotImplementedError("GDN does not support inference for now.")

if packed_seq_params is not None:
# TODO: support packed sequence
raise NotImplementedError("GDN does not support packed sequence for now.")