test: add generated example checks and e2e

[BloggerBust]

Description

This PR adds two test layers for the generated examples. It is stacked on #5.

examples-check is a fast layer that runs on every pull request. It does not start an engine, so it proves the examples compile, not that they run.

examples-e2e is the runtime layer. It installs the Hatchet CLI, starts a real local engine via Docker, and for each example starts the generated worker, triggers the workflow, and asserts the expected output.

The engine tag defaults to latest rather than a pinned version, so the e2e is not yet reproducible against a fixed engine. Pinning is deferred until a tag compatible with the examples' pinned SDKs is chosen. examples-e2e runs on pull_request for now so CI exercises it on this PR. If it proves flaky, it can move to workflow_dispatch and a schedule before merge.

Refs hatchet-dev/hatchet#4107

Type of change

• Test changes (add, refactor, improve or change a test)

What's Changed

• Add the examples-check workflow and hack/check-examples.sh: build and vet Go, install and typecheck TypeScript, install and byte-compile Python
• Add the examples-e2e workflow and hack/e2e-examples.sh: start a real engine, run each generated worker, trigger its workflow, assert the expected output, then remove the workers and containers

Checklist

Changes have been:

• Tested (unit, integration, or manually with steps specified)
• Linted and formatted

Testing

• Ran the e2e locally for all three examples on Node 24, with a CLI built from the sibling hatchet repo. Each started the worker, triggered the workflow, and printed its expected output. Workers and containers were cleaned up.
• The compile-level checks pass: build and vet Go, typecheck TypeScript, byte-compile Python, plus the existing dependency sync and generation drift checks.
• Running TypeScript surfaced a pnpm version bug. The example's pnpm-workspace.yaml uses pnpm 10 settings that pnpm 9 rejects, so both workflows now install pnpm@10, which reads the committed lockfile.
• Not yet exercised, and first run in CI on this PR: the public CLI install script and the pull_request e2e run.

---

🤖 AI Disclosure

• I acknowledge that an LLM was used in the creation of this Pull Request, in accordance with Hatchet's AI_POLICY.md.

• Details: Claude Code wrote the initial test scaffolding.

[mrkaye97]

I didn't look through this one very closely since it's a lot of bash 😅 looks fine to me though, although I wonder if there's a way to simplify more

[BloggerBust]

Hey @mrkaye97, I took another look to see if there was a way to simplify this. I think it is about as simple it can be while still keeping the checks and retries. For now, I am going to merge this as is. Thanks for the review.