Update cluster creation and install script

[nkemnitz]

• system node e2-medium -> e2-standard-2, remove manual CPU request overrides
• keep e2-medium, but enable autoscaling for system nodes (to host the additional kube-dns and konnectivity pods)
• include helm in zutils install (required for KEDA setup when creating new clusters)

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 100.00%. Comparing base (a01d363) to head (22574c1).
Report is 2 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff            @@
##              main      #887   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files          142       142           
  Lines         6113      6113           
=========================================
  Hits          6113      6113           

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

[akhileshh]

Looks good, but any reason to explicitly set --max-pods-per-node "32" ? The default is 110, what are the benefits of limiting it to 32?

[nkemnitz]

That's just a habit from old days where we had thousands of nodes running and would exhaust the IP address ranges, iirc - so actually more relevant to the worker node pools than the system node pool

[supersergiy]

Would it make sense to add more node pools to these clusters, like we have on zetta-research?