Add dummy feature for nrow=1 altExps

[jashapiro]

Closes #24

Here I am adding a check for altExps with only a single feature when converting to a Seurat v5 assay. If there is only one feature, I add second one named null-feature (suggest other names if you have a better idea!) with all zero counts and all NA rowData.

In testing (including a number of added tests), this seems to solve the issue we were having.

(I am assuming that there are not going to be full SCEs with only one feature, or something much bigger is wrong and I am happy to have the function fail.)

[sjspielman]

Everything you've added seems fine to me. The main thing I was thinking about is whether we want to lock this down to be an altExp rather than remain flexible for any SCE including the "main experiment". While in principle this is fine on any SCE, I'm thinking about how other components of the object play out, like reduced dims. Of course with only 1 feature, I can't even imagine a situation where that would be present, let alone metadata like highly-variable features. So, I have now 99.9% talked myself out of thinking about anything besides assays.

This leads to me two comments:

1. We definitely don't want to export this function, and we don't! Done.
2. Can you think of other innards we'd want to check and modify in any way? I really can't, so I'm approving, but if you can let me know!
3. Bonus comment - I guess we could add a check that "null-feature" isn't already the single feature name, but boy would it be wild if it is. I'd have many more questions and would be glad for everything to error out instead.

[sjspielman]

Also want to note that I was pleased to find that opening the Rproj locally didn't try to change the UUID. I didn't expect it to (and as we're both on the new RStudio version, it wasn't deleted either), and I was happy to be right.