Evaluate reference approach for annotating osteosarcoma samples

### If you are filing this issue based on a specific GitHub Discussion, please link to the relevant Discussion.

https://github.com/AlexsLemonade/OpenScPCA-analysis/discussions/1410

### Describe the goals of the changes to the analysis module.

The OsteoCar reference that we will be using to annotate these samples is split across several files:
```
xeno_prim_human_cancer_cells
patient_prim
patient_mets
mm_prim
mm_mets
xeno_prim_mouse
xeno_mets_mouse
dogs_prim
dogs_mets
xeno_mets_human_cancer_cells
```

We have all human samples, including primary, metastatic, and xenograft. A question has arisen about to use the references: We would probably like to consider xenograft and patient references separately, but how different would e.g. SingleR results for patient samples under either approach:
- Run SingleR with two references provided (primary and metastasis)
- Run SingleR with one concatenated reference

We can run this on a few pilot samples and then compare annotations to guide choosing the reference approach. This also seems worth doing: https://bioconductor.org/books/release/SingleRBook/using-multiple-references.html#manual-label-harmonization

### What will your pull request contain?

This will have two PRs:

1. Establish an renv environment and add a script to convert references from Seurat -> SCE. We can save these as individual files, to be concatenated as part of exploration where needed. Depending on that outcome, the conversion script could be modified. Note that I only expect this script to convert to SCE, not to AnnData. The script will be modified later for #1413 once this investigation is complete.
2. An exploratory notebook comparing references and SingleR. I haven't run any of this yet, but it's also possible we get a lot of good insight from just running `SingleR::matchReferences()` as linked above, so that's the starting point for that notebook.

### Will you require additional software beyond what is already in the analysis module?

Yup, renv incoming with SingleR and such

### Will you require different computational resources beyond what the analysis module already uses?

Should be fine on laptop

### If known, when do you expect to file the pull request?

_No response_