perf: reuse V8 template for wall profile context objects

[IlyasShabi]

Summary

V8’s DictionaryTemplate optimizes repeated construction of JS objects with a fixed set of properties. It creates the template once and lets V8 reuse the same object layout for subsequent objects.

During wall-profile collection, we materialize one JS context object for every matched sample. Previously, each object was created empty and populated using up to 4 individual property writes. This adds overhead on both the C++ and V8 sides through repeated property operations and object-shape transitions.

This PR caches a DictionaryTemplate per V8 isolate and uses it to construct these context objects directly with the expected layout.

Benchmark

• Runs wall profiling with contexts, CPU time, and async IDs enabled.
• Measures the native TimeProfiler.stop() phase with process.hrtime.bigint().
• Divides the elapsed time by the number of context objects created.

Time per object on main: 379 ns
Time per object on this PR: 185 ns

This reduces construction cost by more than 50%. The generated objects also go from three V8 object shapes to one consistent shape.

DictionaryTemplate is available starting with Node.js 22

There are other fixed-shape objects in the time and heap profilers that may benefit from the same optimization. I can address those in follow-up PRs if this approach looks good.

[github-actions]

Overall package size

Self size: 2.62 MB
Deduped: 3.32 MB
No deduping: 3.32 MB

Dependency sizes

| name | version | self size | total size | |------|---------|-----------|------------| | pprof-format | 2.3.1 | 504.33 kB | 504.33 kB | | source-map | 0.8.0 | 185.66 kB | 185.66 kB | | node-gyp-build | 4.8.4 | 13.86 kB | 13.86 kB |

_{🤖 This report was automatically generated by heaviest-objects-in-the-universe}

[datadog-datadog-us1-prod]

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: 176e4e5 | Docs | View more details | Give us feedback!}

[szegedi]

This looks great!

I guess the maybe obvious next place to subject to this treatment is the tree builder as the nodes are all also identical dictionaries