Add option to save cached embeddings to disk for re-use

To follow-up #16 and the discussion in [bioclip](https://github.com/Imageomics/bioclip/issues/20#issuecomment-2211021936), create an option for the text embeddings cache to be saved to disk and an option to load these from a specified filepath rather than recomputing if the category list is reused.

Start by reviewing issue #16 and the linked BioCLIP discussion, then locate the text-embedding cache in this repository. The work is complete when cached embeddings can be saved to disk and loaded from a specified filepath for a reused category list without recomputation.

Written by the indexing model from the issue text.