Making the GRUB menu show at every boot with kernel messages instead of the logo, recovery mode, getting a system back from grub rescue>, and rebuilding the initramfs. Tasks 2 and 3 need the VM's console.

Before you start

The lab saves /etc/default/grub and /etc/default/grub.d/ so reset can put them back (and runs update-grub if they changed). It changes nothing else. On cloud images, /etc/default/grub.d/50-cloudimg-settings.cfg is read after /etc/default/grub and sets GRUB_TIMEOUT=0.

Tasks

Make the GRUB menu appear for 10 seconds at every boot, and make every boot show kernel messages instead of the logo (no quiet or splash on the kernel command line). Reboot to see it.
Boot once into recovery mode, get a root shell, and make sure / is writable. Not graded: in that shell, cat /proc/cmdline contains recovery, systemctl is-active friendly-recovery.service prints active, and touch /root/recovery-test succeeds.
Rename /boot/grub to /boot/grub.bak, reboot, and get the system booted from grub rescue>. Then repair it for good. Not graded: after sudo mv /boot/grub.bak /boot/grub, sudo grub-install DISK prints Installation finished. No error reported., and the next reboot shows the normal menu. Take a snapshot of the VM before this one.
Rebuild the initramfs of every installed kernel.

Hints

GRUB_TIMEOUT_STYLE=menu, GRUB_TIMEOUT=10 and GRUB_CMDLINE_LINUX_DEFAULT, in /etc/default/grub or a file in /etc/default/grub.d/ that sorts after the cloud image's; then sudo update-grub. After a reboot, cat /proc/cmdline.
In the menu: Advanced options, then the (recovery mode) entry, then root. mount -o remount,rw / if / is read-only.
At grub rescue>: set shows the prefix GRUB looked in; ls (hd0,gpt16)/ finds grub.bak (on cloud images /boot is its own partition). Then set prefix=(hd0,gpt16)/grub.bak, set root=(hd0,gpt16), insmod normal, normal. lsblk shows the disk for grub-install.
sudo update-initramfs -u -k all prints a Generating line for each kernel in ls /boot/vmlinuz-*.