test: Add Gherkin tests using Behave

[raymyers]

This adds automated tests for every command except dep cycle and migrate-beads. This was part of an experiment porting the command to other languages, and I figured it could be useful to contribute the suite back upstream.

The test format is Gherkin, making them readable as tests and executed using the harness in features/steps/ticket_steps.py.

Closes #8

AI use disclosure

The majority of this was AI generated based on my plan. The change is safe because it's test-only, and I believe I've done a reasonable job ensuring the tests are meaningful. Here was my process:

1. One-shot prompt extracts a SPEC.md from the original ticket script.
2. Agent scans spec for ambiguity or inconsistency with implementation. (original commit 1, commit 2)
3. Agent adapts SPEC into text Gherkin specs. I have read them.
4. Agent adds Behave harness to run the Gherkin specs, and gets them to pass with current ticket script.
5. Agent uses the test suite to create implementations of the CLI in several other languages.
6. Manually spot checked by using the re-implemented CLI to manage some tickets.

Here, "one-shot prompt" means Claude Opus 4.5. "Agent" means OpenHands with Claude Sonnet 4.5, except step 3 which was Claude Code.

[wedow]

Nice! Will be making great use of this as I do some upcoming refactorings. Thanks for putting it together.

Love the AI process walkthrough btw. Great way to do things