fix(cpu_monitor): collect cpu information on timer thread

[nishikawa-masaki]

Description

As reported on #10554 , the "cpu_monitor" node collects the following CPU statistics in the context of publishing the "/diagnostics" topic.

• Load of each core
  • "mpstat -P ALL 1 1 -o JSON" command line is executed.
    • This will be fixed by another PR in near future.
• Frequency of each core
• Temperature of each core
• Overall load average
• Thermal Throttling status about each core (optional)

In addition to "sleeping 1 second in mpstat", other operations take a considerable amount of time.
As a result, publishing intervals from the "cpu_monitor" node are longer than expected (1 second).

This PR makes the following changes to the way the "cpu_monitor" node works:

• "cpu_monitor" creates a timer with 1 second intervals.
• In the context of the timer-callback, the "cpu_monitor" node collects statistics about CPU cores and stores them internally.
• On "update" for publishing, the "cpu_monitor" node publishes the stored statistics information.

Handling of thermal throttling is intentionally left unchanged because:

• The "cpu_monitor" node supports multiple CPU platforms.
• The behaviors of thermal throttling are different among CPU types.
• The structures of thermal throttling information and the procedures for collecting the information are tightly coupled and are different among CPU types.
• Therefore, handling of thermal throttling can't be generalized.
• The load and time for collecting thermal throttling alone won't be a problem as other works are moved to timer-callback context.

Related links

Parent Issue:

• The "cpu_monitor" node takes more than 1 second for publishing while it is expected to publish /diagnostics topic with 1 second intervals. #10554

Private Links:

• TIER IV internal link

How was this PR tested?

• All PR CI tests passed.
  • The source code can be built for different CPU platforms.
  • Not tested on other CPU platforms than Intel, but it should work.
    • Only the context in which statistics are collected has changed.
• The number of entries and their values in "/diagnostics" topic published by the "cpu_monitor" node are same as before.
• Modified publish interval to 0.5 second (2 Hz) temporarily and verified that the "cpu_monitor" node can publish its data with 0.5 second intervals.
  • The version before this PR can't publish with 0.5 second intervals.

Notes for reviewers

None.

Interface changes

None.

The parameters of the "cpu_monitor" node are now explicitly declared as "read-only".
They had been "read-only" (i.e. they can't be modified after initialization) implicitly.

Effects on system behavior

None.

[github-actions]

Thank you for contributing to the Autoware project!

🚧 If your pull request is in progress, switch it to draft mode.

Please ensure:

• You've checked our contribution guidelines.
• Your PR follows our pull request guidelines.
• All required CI checks pass before marking the PR ready for review.

[codecov]

Codecov Report

Attention: Patch coverage is 0% with 275 lines in your changes missing coverage. Please review.

Project coverage is 24.91%. Comparing base (175ca5c) to head (79bd63a).
Report is 6 commits behind head on main.

Files with missing lines	Patch %	Lines
...ystem_monitor/src/cpu_monitor/cpu_monitor_base.cpp	0.00%	237 Missing ⚠️
...ude/system_monitor/cpu_monitor/cpu_information.hpp	0.00%	19 Missing ⚠️
...stem_monitor/src/cpu_monitor/intel_cpu_monitor.cpp	0.00%	19 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #10545      +/-   ##
==========================================
- Coverage   24.96%   24.91%   -0.05%     
==========================================
  Files        1351     1351              
  Lines      105116   105229     +113     
  Branches    39752    39757       +5     
==========================================
- Hits        26238    26219      -19     
- Misses      76386    76519     +133     
+ Partials     2492     2491       -1     

Flag	Coverage Δ		*Carryforward flag
differential	9.36% <0.00%> (?)
total	24.98% <ø> (+0.02%)	⬆️	Carriedforward from fe6d274

*This pull request uses carry forward flags. Click here to find out more.

☔ 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.