* fix: do not set return value for timed out runs
Commands that are killed manually due to timeout rarely returns a
success status code (0). These codes should not be treated as errors
but simply produced because of the timeout.
* fix(windows): use variable to track timeout
Use a variable to track timeout instead of relying on SIGTERM, as
processes on Windows are not killed using signals.
* minor: refactor to make testing easier
* patch: retrieve inputs into object rather than globals
* test: run more "integration" tests in parallel
* test: fix needs and rearrange ci_integration_* jobs
* test: forgot comma
* test: fix sad_path_timeout_minutes assertions
* test: add single ci_all_tests_passed job that can be required for CI rather than each individual job
* test: add single ci_all_tests_passed job that can be required for CI rather than each individual job
* minor: use spawn to stream larger output rather than exec which buffers it
* test: verify distinct error code is returned from large output test
* test: breakout additional integration tests to run in parallel
* test: dont pass/fail PRs for coverage yet
* feat: retry only on specific exit code
* Run ci_cd on all push events
* dedupe step IDs
* add assertions for retry_on_exit_code tests
* minor: implemented suggested fix from @andersfischernielsen
* docs: update readme to reflect new retry_on_exit_code input
Co-authored-by: Anders Fischer-Nielsen <andersfischern@me.com>
