🏃 add input + combined startup benchmarks

[natemoo-re]

Follow-up to #52. Adds new startup benchmarks in mode: walltime to measure the full E2E perf of:

• createInput bootstrap
• combined createTerm + createInput bootstrap inside Promise.all
• render + scan hitting both modules

also sets both benchmark jobs to codspeed-macro runner to avoid hardware variance, which was causing quite a bit of noise

[pkg-pr-new]

Open in StackBlitz

npm i https://pkg.pr.new/clayterm@56

commit: 2fbe488

[codspeed-hq]

Merging this PR will degrade performance by 37.29%

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

❌ 2 (👁 2) regressed benchmarks
✅ 18 untouched benchmarks
🆕 3 new benchmarks

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
🆕	WallTime	createInput	N/A	283 ms	N/A
🆕	WallTime	render + scan (both modules)	N/A	374 ms	N/A
🆕	WallTime	createTerm + createInput	N/A	374.5 ms	N/A
👁	Simulation	long input burst (200 bytes)	979.3 µs	1,265.3 µs	-22.61%
👁	Simulation	printable ASCII (single char)	88.7 µs	174.5 µs	-49.18%

---

_{Comparing bench/combined (2fbe488) with main (ebaf397)}

[cowboyd]

❌ 2 (👁 2) regressed benchmarks

Seeing as this did not actually touch any library code, feels like either we aren't running our benchmarks enough, or we need to run them not in absolute terms but in relative to a control that can account for fluctuations in runtime.

[natemoo-re]

@cowboyd yeah this PR also switched the runners to codspeed-macro which will be consistent—ubuntu-latest varies wildly and was creating a lot of noise. thankfully this should be the last noisy one!