Return ensembl ID for each embedding for emb_mode="gene"

[mattwoodx]

• When "gene" emb_mode is selected, previously we would just return all the raw embeddings. Now we return all gene embeddings in the form of pandas series, where the index of each pandas series is the ensembl ID mapping to the corresponding embeddings
• Updated geneformer testing to follow that of scGPT, allowing for more granular CPU bound unit testing.

[notion-workspace]

Add gene names to embeddings for Geneformer