Deploy existing models using Seldon

[chauhankaranraj]

As a data scientist,
I want to deploy the existing disk failure prediction models as a publicly available inference service,
So that anyone can estimate time to failure for a disk by sending disk health data (e.g. SMART metrics) to a service.

One option to consider is deploying via seldon-core, since the operate-first cluster already has a Seldon operator deployed.

Acceptance Criteria

• discussion with the infra team about using seldon/kfserving/something else
• current models deployed as a service