Pass SystemdCgroup setting through to nvidia runtime options

[brandond]

Proposed Changes

Pass SystemdCgroup setting through to nvidia runtime options.

This is apparently required by newer versions of the nvidia container runtime; something appears to periodically kill off pods using the nvidia runtime if they are not in the correct cgroup.
Ref: NVIDIA/nvidia-container-toolkit#28 (comment)

Types of Changes

bugfix

Verification

see linked issue

Testing

Linked Issues

• pods with runtimeClassName: nvidia get killed after a few seconds #7130

User-Facing Change

Fixed issue that would cause pods using nvidia container runtime to be killed after a few seconds, when using newer versions of nvidia-container-toolkit.

Further Comments

[codecov]

Codecov Report

All modified lines are covered by tests ✅

Comparison is base (2a9e8e6) 47.03% compared to head (5fea74a) 51.39%.
Report is 1 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #8470      +/-   ##
==========================================
+ Coverage   47.03%   51.39%   +4.35%     
==========================================
  Files         143      143              
  Lines       14843    14843              
==========================================
+ Hits         6982     7629     +647     
+ Misses       6740     5995     -745     
- Partials     1121     1219      +98     

Flag	Coverage Δ
e2etests	48.63% <ø> (?)
inttests	44.48% <ø> (+0.10%)	⬆️
unittests	19.73% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

Files	Coverage Δ
pkg/agent/templates/templates_linux.go	62.50% <ø> (ø)

... and 42 files with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.