Fix disk space issues on github runners

[jscheffl]

Main is failing e.g. in https://github.com/apache/airflow/actions/runs/13101805105/job/36550974189 as runners are "out of disk space".
This seems to be caused as only 23GB are free on /root disk where main artifacts like docker is produced. When the runner starts the following disk space is available:

Run df -H
Filesystem      Size  Used Avail Use% Mounted on
/dev/root        78G   56G   23G  72% /
tmpfs           8.4G  177k  8.4G   1% /dev/shm
tmpfs           3.4G  1.2M  3.4G   1% /run
tmpfs           5.3M     0  5.3M   0% /run/lock
/dev/sda15      110M  6.4M  104M   6% /boot/efi
/dev/sdb1        79G  4.3G   71G   6% /mnt
tmpfs           1.7G   13k  1.7G   1% /run/user/1001

(A lot of build tools are already on CI image - see more debug info from the additional debug steps in https://github.com/apache/airflow/actions/runs/13102704519/job/36552932758 where the disk space onthe image is allocated.)

...but this means that after building the image only ~5GB are free and in some corner cases (e.g. main) the build fails. Disk space at then of building pipelines is:

Run df -H
Filesystem      Size  Used Avail Use% Mounted on
/dev/root        78G   73G  5.2G  94% /
tmpfs           8.4G  177k  8.4G   1% /dev/shm
tmpfs           3.4G  1.2M  3.4G   1% /run
tmpfs           5.3M     0  5.3M   0% /run/lock
/dev/sdb15      110M  6.4M  104M   6% /boot/efi
/dev/sda1        79G  8.2G   67G  11% /mnt
tmpfs           1.7G   13k  1.7G   1% /run/user/1001

...and as it seems a LOT of disk space is on the second drive in /mnt and the larged data chunk from triaging is the docker workspace, this PR adds a bind mout to use the second drive in /var/lib/docker. With this at the end of the build of the image the following free space is available:

Run df -H
Filesystem      Size  Used Avail Use% Mounted on
/dev/root        78G   54G   25G  69% /
tmpfs           8.4G  177k  8.4G   1% /dev/shm
tmpfs           3.4G  1.2M  3.4G   1% /run
tmpfs           5.3M     0  5.3M   0% /run/lock
/dev/sda15      110M  6.4M  104M   6% /boot/efi
/dev/sdb1        79G   28G   48G  37% /mnt
tmpfs           1.7G   13k  1.7G   1% /run/user/1001

Adding this as a function to the docker cleanup script and making this to all pipelines.
(Numbers were taken from Python 3.9)

[jscheffl]

Mhm, not sure but smells like image constraint generation failure is un-related to this PR... at least should be...

[amoghrajesh]

Thanks! LGTM +1

[jscheffl]

Docs build problems seem to be un-related, merging. Dcos will need another fix