Use Jinja to manage includes

The drawback to the native `%include` Kickstart directive is that it
requires a static, hard-coded, absolute path.  This means that we
cannot, for example, host a copy of the kickstarts from a different
branch for testing, without modifying the URLs of all the included
files.

Switching to using Jinja templates introduces a build step, but the
result is that the artifacts are self-contained.  This way, they can be
deployed anywhere.  I'm not sure where I'll put them, though, and
they'll need a Jenkins job to run the build and publish them.

fedora-rpi-common.ks.j2

@@ -0,0 +1,30 @@
+{# vim: set ft=sh.jinja : -#}
+bootloader --location none
+clearpart --all --initlabel --disklabel msdos
+part /boot/efi --fstype vfat --size 128 --label RASPBERRYPI
+
+timesource --ntp-pool 2.fedora.pool.ntp.org
+
+%packages --exclude-weakdeps
+bcm2711-firmware
+bcm283x-overlays
+uboot-images-armv8
+%end
+
+{% include "fedora-common.ks" %}
+
+%post --erroronfail
+cp -P /usr/share/uboot/rpi_arm64/u-boot.bin /boot/efi/rpi-u-boot.bin
+
+grubby --args='quiet systemd.show_status=1' --update-kernel=ALL
+
+systemctl enable chrony-wait
+# Ensure clock is synchronized before trying to get host certs signed
+mkdir -p /etc/systemd/system/ssh-host-cert-sign@.service.d
+cat > /etc/systemd/system/ssh-host-cert-sign@.service.d/time-sync.conf <<EOF
+[Unit]
+After=time-sync.target
+Wants=time-sync.target
+Before=sshd.service
+EOF
+%end