issue using cache=True in scanpy read

[bhoellbacher]

Within the additional_qc_scsnake notebook bc.Import.read_mtx() uses scanpy.read with cache=True to read in the .h5ad file.
If the filtering parameters (such as besca_standard_n_genes: 10000 ) were changed after first running the pipeline, the cached file does not get updated.
This leads to a mismatch in the number of observations between the cached .h5ad file and the barcodes.tsv file.

Setting cache=False resolves this issue without impacting loading times in a noticeable fashion.