[FEATURE] Model autotracing workflow for sparse encoding model.

[conggguan]

Is your feature request related to a problem? There have some need to publish sparse encoding pre-trained models. But now opensearch-py-ml only support text embedding models' autotracing workflow.

What solution would you like? Add a workflow for sparse encoding model's auto tracing and release.

What alternatives have you considered?

Do you have any additional context? OpenSearch pre-trained models Neural sparse encoding model v1

[conggguan]

Feel free to assign this issue to me @congguan.