Suggested test: copilot/stop-after-expression-input

## Motivation

github/gh-aw#56983 — `on.stop-after` now accepts a GitHub Actions expression (e.g. `${{ inputs.stop-after }}`) in addition to relative deltas/absolute timestamps, resolved at runtime instead of compile time.

## Proposed test

- **Workflow file**: `test-copilot-stop-after-expression.md`
- **Trigger**: `workflow_dispatch` with a `stop-after` input
- **Engine**: copilot
- **Safe output**: `create-issue` (minimal payload)
- **Variant**: standard

## Minimal test prompt sketch

Add a `workflow_dispatch` input (e.g. `stop_after`) and set frontmatter `on.stop-after: ${{ inputs.stop_after }}`. Dispatch with a past timestamp to assert the workflow's activation check reports "stopped" (agent job skipped, no issue created), and dispatch with a future/absent value to assert normal execution creates the issue.

## New fixtures or secrets needed

None.

## Notes

Compile-time behavior differs from the existing literal `stop-after` support (not previously tested at all in this repo) — this specifically targets the new expression-passthrough path added in gh-aw#56983. Filing as an issue since verifying `gh aw compile` accepts the expression syntax and that the generated lock file correctly gates on the resolved value should be confirmed before locking in the exact prompt/dispatch inputs.

> Generated by [🔍 Suggest New E2E Tests](https://github.com/githubnext/gh-aw-test/actions/runs/33374036874) · copilot · auto · 28.1 AIC · ⌖ 8.6 AIC · ⊞ 8.8K · [◷](https://github.com/search?q=repo%3Agithubnext%2Fgh-aw-test+is%3Aissue+%22gh-aw-workflow-call-id%3A+githubnext%2Fgh-aw-test%2Fsuggest-new-e2e-tests%22&type=issues)

Start with the proposed test-copilot-stop-after-expression.md fixture and run gh aw compile to verify the expression syntax and generated lock file. Add workflow_dispatch coverage for past and future or absent stop-after values, then dispatch both cases and confirm the stopped case skips the agent job while the normal case creates the issue.

Written by the indexing model from the issue text.