osbuild flaking ENOSPC with after build-with-buildah enabled in rawhide

I've noticed a few things flaking runing out of space in the created disk image inside OSBuild

The disk size estimation code seems to be in line with recent non build-with-buildah builds:

link to failed run

08:10:28  Estimating disk size...
08:10:29  Disk sizes: metal: 2839M (estimated), cloud: 10240M

but then we fail in OSBuild:

[2025-08-27T12:15:44.263Z] org.osbuild.copy: e00e83481f729b32f008f561a487906c2de9f4cd0e98e5f1defa6b4d2d0930da {
[2025-08-27T12:15:44.263Z]   "paths": [
[2025-08-27T12:15:44.263Z]     {
[2025-08-27T12:15:44.263Z]       "from": "input://tree/",
[2025-08-27T12:15:44.263Z]       "to": "mount://root/"
[2025-08-27T12:15:44.263Z]     }
[2025-08-27T12:15:44.263Z]   ]
[2025-08-27T12:15:44.263Z] }
[2025-08-27T12:15:44.263Z] device/disk (org.osbuild.loopback): loop0 acquired (locked: False)
[2025-08-27T12:15:44.263Z] mount/root (org.osbuild.xfs): mounting /dev/loop0p4 -> /home/jenkins/agent/workspace/test-override/cache/osbuild/store/tmp/buildroot-tmp-u0z210wp/mounts/
[2025-08-27T12:15:44.516Z] mount/boot (org.osbuild.ext4): mounting /dev/loop0p3 -> /home/jenkins/agent/workspace/test-override/cache/osbuild/store/tmp/buildroot-tmp-u0z210wp/mounts/boot
[2025-08-27T12:15:44.516Z] mount/efi (org.osbuild.fat): mounting /dev/loop0p2 -> /home/jenkins/agent/workspace/test-override/cache/osbuild/store/tmp/buildroot-tmp-u0z210wp/mounts/boot/efi
[2025-08-27T12:15:44.516Z] Failed to open file "/sys/fs/selinux/checkreqprot": Read-only file system
[2025-08-27T12:15:44.769Z] copying '/run/osbuild/inputs/tree/.' -> '/run/osbuild/mounts/.'
[2025-08-27T12:15:51.285Z] 
[2025-08-27T12:15:51.285Z] Filesystem      Size  Used Avail Use% Mounted on
[2025-08-27T12:15:51.285Z] /dev/vdb1        45G  9.5G   36G  22% /home/jenkins/agent/workspace/test-override/cache
[2025-08-27T12:15:51.285Z] 
[2025-08-27T12:16:01.203Z] cp: cannot create symbolic link '/run/osbuild/mounts/././ostree/repo/objects/68/2244a3557ca80c49194eb34ea0270ee803ccea61a4c623c07e3b8da77934e6.file': No space left on device

log.txt

A subsequent run passes in this step. I'm not sure what is going on here.

Start with the linked Jenkins failed run and log.txt, then compare the failing and subsequent passing runs. Trace the disk size estimation and the OSBuild step with build-with-buildah enabled, focusing on the reported ENOSPC during the ostree copy. Done means identifying the cause of the flake and preventing the created image from running out of space.