recommended/default data type for visualisation is obscure

I think the docs for this can be improved, particularly in the "visualising marker genes" vignette and in the sc.pl.xxx api documentation. I noticed the following:

(1) The "use_raw" parameter documentation for e.g. sc.pl.dotplot does not make it clear that .raw is used by default if present. It seems this issue is addressed upstream but not picked up for the html docs yet. Please compare: https://scanpy.readthedocs.io/en/stable/generated/scanpy.pl.dotplot.html with https://github.com/theislab/scanpy/blob/9360422823fade22a625f3c7840bef132364027e/scanpy/plotting/_anndata.py#L109

(2) It would be very helpful to provide and link in the docs to the script used to pre-process/create the example datasets such as sc.datasets.pbmc68k_reduced().

(3) The visualisation vignette (https://scanpy-tutorials.readthedocs.io/en/multiomics/visualizing-marker-genes.html) does not state that normalised log1p data is being used for the example visualisations (and is presumably recommended). By itself the warning against use of scaled data leaves the situation ambiguous.

Because of these issues, and apologies if I missed the explanation, it is currently not straightforward to be sure what data is being plotted in the dot plots shown in this vignette. From inspecting the anndata object, reading the _anndata.py code and reading the 3k vignette I assume that it is the log1p normalised data!

Many thanks, S.

Start with the visualising marker genes vignette and the sc.pl.dotplot API documentation, comparing the rendered docs with scanpy/plotting/_anndata.py around line 109. Review sc.datasets.pbmc68k_reduced() and identify whether the example uses log1p-normalised data. Done means the docs explain the default use_raw behavior, clarify the plotted data, and link the example-dataset preprocessing script if available.