Show CPU/memory usage in Qube manager again

[alimirjamali]

fixes: QubesOS/qubes-issues#7817

[codecov]

Codecov Report

❌ Patch coverage is 83.33333% with 7 lines in your changes missing coverage. Please review.
✅ Project coverage is 69.19%. Comparing base (8dd162e) to head (4301eb1).
⚠️ Report is 39 commits behind head on main.

Files with missing lines	Patch %	Lines
qubesmanager/qube_manager.py	89.74%	4 Missing ⚠️
qubesmanager/utils.py	0.00%	3 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #424      +/-   ##
==========================================
+ Coverage   69.05%   69.19%   +0.13%     
==========================================
  Files          17       17              
  Lines        3855     3895      +40     
==========================================
+ Hits         2662     2695      +33     
- Misses       1193     1200       +7     

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[marmarta]

I played around with this PR for a bit, and while I generally love it, I have comments:

• sorting: I think that sorting on usage is kinda problematic right now; it keeps dom0 as top item, which is a bit counterintuitive but consistent with the rest of the code, but also on ascending sorting puts all the non-running VMs on top and the running ones on bottom, which is definitely unintuitive, and on descending sorting puts dom0 at the very end, even though it is definitely running. I think any sorting on usage should put VMs that are consuming resources on top, disregarding special dom0 status too.
• I get different numbers on CPU usage in qube manager and in domains widget, which is not great; there seems to be different rounding, because the error is off-by-one sometimes, but I think they need to be consistent

[alimirjamali]

I played around with this PR for a bit, and while I generally love it, I have comments:

* sorting: I think that sorting on usage is kinda problematic right now; it keeps dom0 as top item, which is a bit counterintuitive but consistent with the rest of the code, but also on ascending sorting puts all the non-running VMs on top and the running ones on bottom, which is definitely unintuitive, and on descending sorting puts dom0 at the very end, even though it is definitely running. I think any sorting on usage should put VMs that are consuming resources on top, disregarding special dom0 status too.

* I get different numbers on CPU usage in qube manager and in domains widget, which is not great; there seems to be different rounding, because the error is off-by-one sometimes, but I think they need to be consistent

The suggested changes will be applied. I changed the PR status to draft while I work on it and until I finish it.

[Sahilll10]

Hi @alimirjamali @marmarta
I am working on the System Health Monitor project for GSoC 2026, and in order to do some groundwork before the Pre-GSOC period, I wanted to ask if I can pick up the sorting fix that marmarta flagged.

After thoroughly going through the files and code, the sorting for CPU and MEM columns currently returns int(vm.CPU_usage) directly.

The problem is:
Non-running VMs return None for CPU_usage so they end up sorted incorrectly above running VMs in ascending order
dom0 has no special handling so it falls to the bottom in descending order

The fix I am thinking of — 1) for CPU and MEM sort values, return -1 for non-running VMs so they always sort below running ones, and keep dom0 pinned at the top the same way it is handled in the existing UserRole + 1 sort logic.
2) For the rounding difference between qube_manager and domains widget — domains widget uses int(cpu_usage) directly from vm-stats.
The PR rounds the same way so I need to check where the off-by-one comes from.

Would it be okay if I open a PR with this fix building on your work? I will be testing it using MockQubesComplete.