Bug/save pca model gpumap

[Ben-Epstein]

we need to save the pca data so that we can properly run inference. Inference needs the PCA data from training in order to get its embeddings in the same latent space, both for visualization and drift detection

Tests

Installed from this branch, and did a run

run

And see the pca model in minio

data

Then i did an inference run

and the embeddings look correctly aligned