Lint and validate generated output files in tests

[kdeldycke]

Problem

Generated output files (thin-caller workflows, config merges, changelog markdown) are assembled via string building in Python. Structural bugs — like the Python dict literal leak fixed in the _render_trigger_value() regression — are only caught by pattern-matching assertions, not by the same tools that lint these files in CI.

Proposed solution

Add post-generation validation in tests using the project's existing linting tools:

High priority (files written to disk in downstream repos)

Generated output	Source	Validation tool
Thin-caller workflows	workflow_sync.generate_thin_caller()	yamllint + actionlint
Workflow headers	workflow_sync.generate_workflow_header()	yamllint + actionlint
Labeller rule files	init_project._init_config_files()	yamllint
TOML config merge into pyproject.toml	init_project.init_config()	ruff format (TOML)
Changelog markdown	changelog.py render + write	mdformat

Medium priority (ephemeral or stdout-only)

Generated output	Source	Validation tool
Native TOML configs from [tool.X] bridge	tool_runner._dict_to_toml()	structural TOML validation
Native JSON configs from [tool.X] bridge	tool_runner.resolve_config()	jq
PR/issue body markdown	pr_body.render_template()	mdformat

Context

test_no_python_literals_in_yaml was added as a regression test after thin-caller generation serialized workflow_dispatch input definitions as Python dict literals ({'description': '...'}) instead of block-style YAML. The test catches that specific pattern, but running yamllint and actionlint on the generated output would catch a broader class of structural issues.