IRIIIF should avoid unnecessary loading and processing.

[kaladay]

The AbstractManifestController.java shows a request, a manigest load, and then an I/O copy of the data.

The Dspace Manifest collect performs this specific task.

The Fedora Manifest collect performs this specific task.

This is extra unnecessary overhead that can become notable for larger manifests.

Doing this would likely improve performance and may help resolve converns from the Cantaloupe Response Time Issue. This obviously will not fix performance problems directly in Cantaloupe but should reduce the additional overhead of copying and then transfering.

This also relates to and is likely affected by the #146 story. (Such as using Ingress or using a long-term-cache.)

Possible implementations might include using HTTP HEAD request to determine existence and then performing a fetch to the appropriate server.
If this can be done automatically via an Ingress, then that would be even better.
Example here:

• https://kubernetes.io/docs/concepts/services-networking/ingress/

The following Ingress tells the backing load balancer to route requests based on the Host header. ...