Show the species-wise accuracy in the machine prediction component

[mihow]

Summary

We show the confidence score for each prediction, however the model performs differently for each species that it was trained on. The accuracy for each species is calculated after a model is trained, but this data is not available for anyone to see. Show this information to the user so they can better interpret the validity of the prediction on their own.

Implementation Details

• Store the species-wise accuracy in a publicly accessible place (json in the object store, next to the category map json). Add to model class in ami-data-companion
• Report species wise accuracy from the ML backend with the algorithm category map (ami-data-companion)
• Add new field(s) to AlgorithmCategoryMap model in Antenna
• Return accuracy with the classification API response for the UI
• Update machine prediction component

We could additionally show the species-wise accuracy on the taxon detail pages. However we would need to show the score from each model that can predict this species (which would be very helpful to see). Similar to this card on iNaturalist, but it would need to support multiple models. This is likely its own ticket / task.