feat(build): Add benchmarks framework #10

	name: Benchmarks

	# CodSpeed uses Valgrind to count CPU instructions.
	# Valgrind wraps `hatch run` and traces child processes (hatch → uv → python → pytest).
	# Do NOT skip hatch/uv in Valgrind args. Skipping a parent process causes Valgrind to lose track of ALL children (no data captured).
	# See `benchmarks/README.md` for more information.

	on:
	pull_request:
	push:
	branches: [main]

	permissions:
	contents: read
	id-token: write

	jobs:
	benchmark-fast:
	runs-on: ubuntu-latest
	timeout-minutes: 30

	steps:
	- uses: actions/checkout@v4
	with:
	fetch-depth: 0 # Required for baseline comparison

	- name: Set up Python
	uses: actions/setup-python@v5
	with:
	python-version: "3.11"

	- name: Install hatch
	run: pip install hatch

	- name: Run benchmarks with CodSpeed
	uses: CodSpeedHQ/action@v4
	env:
	UV_NO_PROGRESS: "1"
	with:
	run: hatch run benchmark:run --codspeed -m "not slow"

	- name: Generate benchmark JSON (fallback)
	if: always()
	env:
	UV_NO_PROGRESS: "1"
	run: \|
	hatch run benchmark:run \
	--benchmark-only \
	--benchmark-json=benchmark-results.json \
	-m "not slow" \|\| true

	- name: Upload benchmark results
	uses: actions/upload-artifact@v4
	if: always()
	with:
	name: benchmark-results
	path: benchmark-results.json
	retention-days: 30

Provide feedback