feat: Adding Dockerfile for stats and CI for building container images

[malteos]

This PR adds a Dockerfile for the stats workflow including a GitHub action that runs the unit tests, builds the images (stats and site), and pushes the images to the GitHub registry (fix for #20).

The new plots were generated with the container run command from the README but with the image from the feature branch:

docker pull ghcr.io/malteos/cc-crawl-statistics/stats:feat-docker-ci

[damian0815]

I notice that the background color of the images is transparent - is that intended?

[damian0815]

otherwise,, very nice

[malteos]

The transparent background is coming from the ggplot minimal theme (probably due to a new version)

See #23 (comment)

Since the background on the website (https://commoncrawl.github.io/cc-crawl-statistics/plots/tld/by-year-and-continent.html) is white, it doesn't matter.
I have no explanation for it, except a different ggplot2 version used to generate the plot.

Or tidyverse/ggplot2#4919 (comment)

I tried to set the background explicitly but it doesn't work :(

[malteos]

There are still minor differences (font size and colors in url status by year).

[malteos]

Font size, line widths and colors of url status by year are now adjusted. Only plots/tld/groups.png has a different color assignment but this should not be relevant (probably due to inconsistent ordering of dict keys).

[malteos]

Only plots/tld/groups.png has a different color assignment but this should not be relevant (probably due to inconsistent ordering of dict keys).

@sebastian-nagel is this critical or can we just merge this?

[sebastian-nagel]

is this critical or can we just merge this?

No. It's not critical.