Handling of case-variant column names in CosMx data

When reading in CosMx data, the metadata file read in to create the obs attribute of the anndata object that becomes the table of the SpatialData object can include both a "cell_id" column and a "cell_ID" column (with different values). This causes the column name sanitization routine in validation to throw an error during SpatialData object creation in the CosMx reader. One way of handling this is to rename the "cell_id" column, which allows the table to validate and the SpatialData object to instantiate.

I have a branch of a fork of this repo that has this behavior implemented and can submit a pull request unless another solution is preferred.

Start with the CosMx reader and the validation column-name sanitization routine described in the issue. Reproduce a read using metadata containing both "cell_id" and "cell_ID", then verify that the resulting SpatialData object instantiates successfully without a validation error.