Fixup smoke test for FindClusters

[dcollins15]

Updates the smoke test in test_find_clusters.R to accommodate variability in label assignments given by FindClusters across different systems.

On 14/01/2024, the maintainers of Seurat were contacted by CRAN with the following message:

---

Dear maintainer,

Please see the problems shown on
https://cran.r-project.org/web/checks/check_results_Seurat.html.

Please correct before 2025-01-28 to safely retain your package on CRAN.

There is a check service for M1mac issues: see
https://www.stats.ox.ac.uk/pub/bdr/M1mac/README.txt .
However, it is running a much older version of the OS
and toolchain -- and the latter often matters.

The CRAN Team

---

tl;dr—we have until 2025/01/28 to resolve the M1mac errors currently being thrown by the CRAN package checks for Seurat:

Failure (test_find_clusters.R:39:3): Smoke test for `FindClusters`
results[[1]] not equal to factor(3, levels = 0:5).
1 string mismatch

Failure (test_find_clusters.R:40:3): Smoke test for `FindClusters`
results[[15]] not equal to factor(4, levels = 0:5).
1 string mismatch

Failure (test_find_clusters.R:41:3): Smoke test for `FindClusters`
results[[24]] not equal to factor(0, levels = 0:5).
1 string mismatch

I'm not totally sure what's causing the difference but I was able to reproduce the error on my local machine (M3mac) after adjusting my environment some. If we're unhappy with this variability I can try to investigate further but for now, I'm pretty confident that this update will resolve the issue. The fix is simple enough, instead of spot-checking the cluster labels for specific cells, the test now checks:

• That every cell was assigned a cluster label (no null values)
• That the correct cluster labels were assigned (but not to which cells)
• That the distribution of cluster sizes matches the expected output

Do we think it's worth introducing a tolerance onto the cluster size check? 🤔

Once it has been approved and merged I will immediately start the process of releasing Seurat 5.2.1 to CRAN 🚀