[autonomy-atomicity] Shared GraphQL test stub couples two review-thread scripts

## Autonomy / Atomicity Findings

### 1. Shared test stub forces unrelated workflow changes through one file
**Category:** Over-broad test
**File(s):**
- `tests/test_review_thread_pagination.py`
- `claude-workflows/pr-review/scripts/pr-existing-comments.sh`
- `claude-workflows/mention-in-pr/scripts/gh-get-review-threads.sh`

**Problem:**
`tests/test_review_thread_pagination.py` validates two different scripts from different workflow areas in one test module, using a single shared `gh` stub generator (`_write_gh_stub`). That stub contains query-shape branching used by both tests (`tests/test_review_thread_pagination.py:13-147`, especially `:53-55` and `:115-139`).

At the same time, the two scripts request different GraphQL shapes:
- `pr-existing-comments.sh` requests `originalCommit` (`claude-workflows/pr-review/scripts/pr-existing-comments.sh:93-94` and `:134-135`)
- `gh-get-review-threads.sh` does not (`claude-workflows/mention-in-pr/scripts/gh-get-review-threads.sh:50-56` and `:87-91`)

Because both tests depend on one branching stub, a legitimate query/field change for one script often requires touching this same shared fixture and can break the other test path. In parallel development, this creates avoidable merge conflicts and cross-feature red builds in a file unrelated to the second change.

**Suggested fix:**
Split this into per-script fixtures/tests:
1. Keep one test module focused on `pr-existing-comments.sh` with a stub fixture that explicitly includes commit metadata.
2. Keep another module focused on `gh-get-review-threads.sh` with only its required query fields.
3. Share only minimal generic helpers (e.g., `parse_fields`) while separating script-specific response payloads.

## Suggested Actions
- [ ] Split `tests/test_review_thread_pagination.py` into script-specific test modules and fixtures.
- [ ] Remove query-shape branching from a single shared stub so one script’s GraphQL changes do not require edits to the other script’s test path.
- [ ] Keep pagination assertions isolated per script to reduce cross-feature failures during concurrent development.

---
[What is this?](https://ela.st/github-ai-tools) | [From workflow: Trigger Autonomy Atomicity Analyzer](https://github.com/elastic/ai-github-actions/actions/runs/30930210940)

Give us feedback! React with 🚀 if perfect, 👍 if helpful, 👎 if not.

Start with tests/test_review_thread_pagination.py, especially _write_gh_stub and the pagination tests, then compare the GraphQL requests in pr-existing-comments.sh and gh-get-review-threads.sh. Run the existing review-thread pagination tests before separating them into script-specific modules and fixtures. Done means each script has isolated query-shape data and pagination assertions, with only minimal generic helpers shared.

Written by the indexing model from the issue text.