docs: embedd asciinema casts (#1154)

Signed-off-by: Fabian Kammel <fk@edgeless.systems>
Co-authored-by: Moritz Eckert <m1gh7ym0@gmail.com>
Co-authored-by: Thomas Tendyck <tt@edgeless.systems>
Co-authored-by: 3u13r <lc@edgeless.systems>

docs/screencasts/generate-readme-svg.sh

@@ -0,0 +1,32 @@
+#!/usr/bin/env bash
+#
+# This script prepares the environment for expect scripts to be recorded in,
+# executes all scripts, and copies the .cast files to our doc's asset folder.
+#
+# Note: A cluster is created in GCP. Therefore you are expected to be logged in
+# via `gcloud` CLI. You credentials at $HOME/.config/gcloud are mounted into the
+# screenrecordings container. A full script run takes ~20min.
+#
+
+docker build -t screenrecodings docker
+
+# Create cast
+docker run -it \
+  -v "${HOME}"/.config/gcloud:/root/.config/gcloud \
+  -v "$(pwd)"/recordings:/recordings \
+  -v "$(pwd)"/constellation:/constellation \
+  screenrecodings /scripts/github-readme.expect
+
+# Fix meta data: width and height are always zero in Docker produced cast files.
+# Header is the first lint of cast file in JSON format, we read, fix and write it.
+head recordings/github-readme.cast -n 1 | yq e -M '.width = 95 | .height = 17' - > new_header.cast
+# Then append everything, expect first line from original cast file.
+tail -n+2 recordings/github-readme.cast >> new_header.cast
+
+# Then render cast into svg using:
+#   https://github.com/nbedos/termtosvg
+termtosvg render new_header.cast readme.svg -t window-frame.svg
+
+# Copy and cleanup
+cp readme.svg ../static/img/shell-windowframe.svg
+rm readme.svg new_header.cast