Fix access to layer data for DietSeurat in V5.

[ddiez]

This fixes #8054 which is caused by a change in satijalab/seurat-object@d6aa9af that changed the behavior of brakets.

[ddiez]

I want to nudge this PR given that it has been a long time since submitted, several patches to Seurat V5 have been released and recently version 5.1.0 came out without a fix, and people are still having problems as indicated by all the imaginative ways used to overcome this in #8054. Tagging @mojaveazure since the issue was caused by a change in SeuratObject. This shows the problem with the latest version of Seurat. Using DietSeurat on pbmc_small dataset drops the dimensionality reduction objects but all the data objects remain.

> pbmc_small
An object of class Seurat
230 features across 80 samples within 1 assay
Active assay: RNA (230 features, 20 variable features)
 3 layers present: counts, data, scale.data
 2 dimensional reductions calculated: pca, tsne

> DietSeurat(pbmc_small)
An object of class Seurat
230 features across 80 samples within 1 assay
Active assay: RNA (230 features, 20 variable features)
 3 layers present: counts, data, scale.data

Specifying which layers to remove also does not work:

> DietSeurat(pbmc_small, layers="counts")
An object of class Seurat
230 features across 80 samples within 1 assay
Active assay: RNA (230 features, 20 variable features)
 3 layers present: counts, data, scale.data

Internally DietSeurat tries to remove the layers using object[[assay]][[lyr]] <- NULL but this does not work and there is no error, the request is silently ignored:

> pbmc_small[["RNA"]][["scale.data"]] <- NULL
> pbmc_small
An object of class Seurat
230 features across 80 samples within 1 assay
Active assay: RNA (230 features, 20 variable features)
 3 layers present: counts, data, scale.data
 2 dimensional reductions calculated: pca, tsne

The right way to accomplish this given the changes in SeuratObject is to use single brakets:

> pbmc_small[["RNA"]]["scale.data"] <- NULL
> pbmc_small
An object of class Seurat
230 features across 80 samples within 1 assay
Active assay: RNA (230 features, 20 variable features)
 2 layers present: counts, data
 2 dimensional reductions calculated: pca, tsne

I hope this helps implementing a fix into the main tree. Perhaps there is some reason I cannot see that prevents this fix from working. In that case I apologize. Just wanted to get this from my mind if possible!