fix(dtypes,strings): never abort cleaning on an unusual object cell #424

	name: Reproducibility gate

	# Fast PR-CI check: the committed CleanBench result is still valid (schema,
	# dataset hashes, environment metadata, release-gate consistency) and every
	# README trust claim still resolves to its backing test/doc/benchmark. Does
	# NOT re-run the full benchmark suite — that is benchmark.yml / cleanbench-full.yml
	# on a nightly/release-candidate cadence.
	on:
	pull_request:
	push:
	branches: [main]

	permissions:
	contents: read

	jobs:
	reproducibility:
	runs-on: ubuntu-latest
	timeout-minutes: 10
	steps:
	- uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
	- uses: actions/setup-python@5fda3b95a4ea91299a34e894583c3862153e4b97 # v7.0.0
	with:
	python-version: "3.12"
	- name: Install
	run: \|
	python -m pip install -U pip
	pip install -e ".[bench,cli]"
	- name: Verify committed CleanBench result
	run: python -m benchmarks.cleanbench --verify-results benchmarks/cleanbench/results/latest.json
	- name: Audit README trust claims
	run: python -m benchmarks.cleanbench.reproducibility audit-readme

Navigation Menu