feat(profiling): internal metrics for overhead

[morrisonlevi]

Description

This is an alternative to #3591. Instead of putting the time we spend stack walking into the timeline and flamegraph, we instead aggregate it into counters, which we emit into internal metrics. This addresses some of the concerns I received about that PR while still providing something we believe is useful: the ability to understand our overhead for a specific profile. This internal metric could be re-exported in the future if we desire to do per-organization or other type of aggregation, for example, to determine which customers are hitting the highest amounts of overhead.

In addition to collecting CPU time walking the stack, this also aggregates CPU time spent in our two background threads, ddprof_time and ddprof_upload. The former is a bit of a misnomer because it also aggregates CPU samples, it doesn't just track time.

Manually verified it's working (practically idle service), this is on a slightly older version of the PR so it looks a little different now:

Reviewer checklist

• Test coverage seems ok.
• Appropriate labels assigned.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 62.11%. Comparing base (32aaf0a) to head (76483bb).

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #3616      +/-   ##
==========================================
- Coverage   62.21%   62.11%   -0.11%     
==========================================
  Files         141      141              
  Lines       13387    13387              
  Branches     1753     1753              
==========================================
- Hits         8329     8315      -14     
- Misses       4260     4273      +13     
- Partials      798      799       +1     

see 3 files with indirect coverage changes

---

Continue to review full report in Codecov by Sentry.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 32aaf0a...76483bb. Read the comment docs.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[pr-commenter]

Benchmarks [ profiler ]

Benchmark execution time: 2026-02-03 19:57:47

Comparing candidate commit aed7a13 in PR branch levi/stack-walk-metric with baseline commit 32aaf0a in branch master.

Found 2 performance improvements and 3 performance regressions! Performance is the same for 24 metrics, 7 unstable metrics.

scenario:php-profiler-timeline-memory-with-profiler

• 🟥 execution_time [+54.337ms; +75.399ms] or [+5.855%; +8.124%]
• 🟩 cpu_usage_percentage [-4.885%; -3.086%]

scenario:php-profiler-timeline-memory-with-profiler-and-timeline

• 🟥 cpu_system_time [+38.635ms; +61.757ms] or [+7.237%; +11.568%]
• 🟥 execution_time [+93.618ms; +115.961ms] or [+7.583%; +9.393%]
• 🟩 cpu_usage_percentage [-6.534%; -5.345%]

[realFlowControl]

I was reviewing this and made a few additions that build on your work in #3618, let me know what you think