Feature Request: RLE Plots

Is your feature request related to a problem? Please describe.

Right now there is no method that plots an RLE plot even though the DeseqDataSet method has both the normalized counts (DeseqDataSet.layers["norm_counts"]) and estimated size factors (DeseqDataSet.obsm["size_factors"]). RLE plots are useful for identifying technical variation and normalization issues in RNA-Seq data, as well as testing out different normalization strategies. These plots are super helpful in assessing and visualizing unwanted technical noise or batch effects in expression data.

Describe the solution you'd like
I propose implementing an RLE plot method, similar to the plotRLE function in EDASeq for R. The plot would display boxplots of the relative log expression of genes for each sample, centered around the median per gene. The implementation would:

• Compute the median for each gene across all samples.
• Calculate the log expression values relative to this median.
• Generate a boxplot for each sample based on relative log expression values.

Describe alternatives you've considered

The implementation in other standard libraries (pandas, numpy, matplotlib) is not too bad, but it would be super handy if it came standard!

Additional context
Here's an example of an RLE plot I've made that could be similar. Thanks!!

Start by inspecting the DeseqDataSet entry point and its norm_counts layer and size_factors obsm data, then compare the requested behavior with EDASeq's plotRLE function. Done means providing an RLE plotting method that computes per-gene medians, relative log expression, and one boxplot per sample.