CI tests fail with `[Errno 2] No such file or directory: '/var/cache/dnf/metadata_lock.pid'`

[vashirov]

Issue Description
There is an intermittent issue during container startup in our CI:

Status: Downloaded newer image for quay.io/389ds/ci-images:test
+ CID=cff2da4e43a2a712df774dbbe8b99206791e8592d6328261965b6f802140bdb7
+ sudo docker exec cff2da4e43a2a712df774dbbe8b99206791e8592d6328261965b6f802140bdb7 sh -c 'dnf install -y -v dist/rpms/*rpm'
Loaded plugins: builddep, changelog, config-manager, copr, debug, debuginfo-install, download, generate_completion_cache, groups-manager, needs-restarting, notify-packagekit, playground, repoclosure, repodiff, repograph, repomanage, reposync, system-upgrade
DNF version: 4.19.2
cachedir: /var/cache/dnf
User-Agent: constructed: 'libdnf (Fedora Linux 40; container; Linux.x86_64)'
repo: downloading from remote: copr:copr.fedorainfracloud.org:group_3[89](https://github.com/389ds/389-ds-base/actions/runs/9280063355/job/25534394245?pr=6182#step:7:90)ds:perl-Mozilla-LDAP
Copr repo for perl-Mozilla-LDAP owned by @389ds 8.4 kB/s | 1.6 kB     00:00    
copr:copr.fedorainfracloud.org:group_389ds:perl-Mozilla-LDAP: using metadata from Mon Feb 12 14:44:12 2024.
repo: downloading from remote: fedora
countme: no event for fedora: window already counted
Fedora 40 - x86_64                              6.9 MB/s |  20 MB     00:02    
fedora: using metadata from Sun Apr 14 18:51:11 2024.
repo: downloading from remote: updates
countme: no event for updates: window already counted
Fedora 40 - x86_64 - Updates                    3.4 MB/s | 6.6 MB     00:01    
updates: using metadata from Tue May 28 01:05:10 2024.
[Errno 2] No such file or directory: '/var/cache/dnf/metadata_lock.pid'
Error: Process completed with exit code 1.

With the recent container image update to F40 these failures are happening almost all the time.

The race condition happens here:

389-ds-base/.github/workflows/pytest.yml

Lines 92 to 93 in c019af1

	CID=$(sudo docker run -d -h server.example.com --ulimit core=-1 --cap-add=SYS_PTRACE --privileged --rm --shm-size=4gb -v ${PWD}:/workspace quay.io/389ds/ci-images:test)
	sudo docker exec $CID sh -c "dnf install -y -v dist/rpms/*rpm"

systemd is not fully initialized when a second command runs and expects a pid file to be available.
The script should wait until systemctl is-system-running is successful.

[vashirov]

be90794..1e10a93 389-ds-base-3.0 -> 389-ds-base-3.0
f48026f..7a23a4f 389-ds-base-2.5 -> 389-ds-base-2.5
e4c1e7e..6cb838b 389-ds-base-2.4 -> 389-ds-base-2.4
90156ca..eec80a9 389-ds-base-2.3 -> 389-ds-base-2.3
ab44858..5430b3f 389-ds-base-2.2 -> 389-ds-base-2.2
adc92a9..aff49bf 389-ds-base-1.4.3 -> 389-ds-base-1.4.3