WIP: create reports with python

[danilexn]

This PR will replace the generation of reports with R in favor of Python (to reduce the dependencies).

It adds a new spacemake/report folder, with the scripts qc_sequencing.py and automated_analysis.py. These take the same variables as the original Rmd notebooks, and generate very similar plots/tables.

jinja2 is used for HTML template rendering. Templates are contained under spacemake/report/templates.

This also addresses #75 by showing the table of all mapping statistics following the sample's mapping strategy, in the same order (from more to less input reads, from top to bottom).

TODO:

• Implement + test saturation analysis report generation.
• Run full pipeline tests on various spatial (non-meshed), spatial (meshed), non-spatial datasets.
• Run full pipeline tests on downsampling analysis (when implemented)
• Separate the ... and footer scripts of the HTML templates into separate files to avoid redundant code
• Implement interactive plotting with plotly/datashader (large sts datasets)

[danilexn]

For the spatial qc plots, we rely on scanpy.pl.spatial. There's no documented easy way of changing the plot marker to arbitrary shapes (e.g., hexagons). We might consider implementing our own code for this.

[danilexn]

Also, I have replaced the previous Nucleotide distribution per beads with something more similar to what FastQC generates (line plots instead of barplots). We can go back to the format in the previous Rmd, it just does not look so good with matplotlib out of the box.

[danilexn]

This PR should also address #57. If a log file that is required has size 0, we can throw one of spacemake's errors.