Eljas Roellin

CONSORT Flowchart background: https://www.bmj.com/content/340/bmj.c332 https://jamanetwork.com/journals/jama/fullarticle/2799401

A few points on this: - I wonder if it is a good idea to store computed results (e.g. `adata.obsm["X_pca"]`) together with altered data that would actually yield different results...

Thanks for the example - so to my understanding you add some (maybe externally computed) statistics to the `.obs` field, and then move these to the `.X` field. Quick clarification,...

Thanks for the diagram, I think I get the point here - it is data in the .obsm in your case, not a computed result. The .obsm field in the...

That would be `.obsm`, `.uns`, `.obsp` you'd consider taking in?

I'd be in favor of not having these values copied now, but storing this data later more consistently. But I'm also fine having options for copying .obsm, .uns, .obsp (while...

I'd close this PR for now, as I think the underlying outline of data handling patterns will not be the way to go - the .obsm field is likely not...

> Please deduplicate the tests though, they have too many identical lines. To do so did across-setting tests with for loop on top of former test... Would you prefer one...

To test in the future that this bug doesnt happen, the 4 combinations of inplace=True/False and subset=True/False need to be compared for their selected `var_names` - if these arguments were...

Timeout and Scrublet failing in Python3.12?