Initial methods to skip already-processed images.

[mihow]

• Adds method to store and check what images and what detections have been processed by what algorithms
• Updates the ML backend schema to nest detections and classifications under the source image response, to more readability associate related objects
• Add new tests
• Fix related bugs & add new logging related to these features

Future work:

• Populate the algorithm metadata to all existing classifications so they won't be reprocessed
• Add support for reusing detections if only the classification algorithm has changed
• Reduce logging related to this change

IMPORTANT: the current ML backend is the "api-only-prototype" branch in the ami-data-companion repo. Launch with the ami api command. This is running on gpu-4 in the Arbutus cloud.
RolnickLab/ami-data-companion#56

[netlify]

✅ Deploy Preview for ami-storybook canceled.

Name	Link
🔨 Latest commit	b3031bb
🔍 Latest deploy log	https://app.netlify.com/sites/ami-storybook/deploys/660b4125fbf5a100086231b3

[netlify]

✅ Deploy Preview for ami-web canceled.

Name	Link
🔨 Latest commit	b3031bb
🔍 Latest deploy log	https://app.netlify.com/sites/ami-web/deploys/660b4125f46b6b0008a07f27