[product-manager-impersonator] Quick setup retry/backoff for workflow downloads

## Feature Idea

**Summary:** Add built-in retry/backoff and timeout handling to `scripts/quick-setup.sh` download steps so transient network failures don’t abort setup.

## Why a Customer Would Want This
Teams commonly run the one-line quick setup command in CI-like or corporate network environments where transient DNS/TLS/429 errors happen. Right now a single failed `curl` during workflow fetch aborts the whole run, even after branch checkout has started. That creates avoidable reruns and slows first-time adoption.

## Rough Implementation Sketch
- Add a small `download_with_retry()` helper in `scripts/quick-setup.sh` that wraps `curl` with retry/backoff and sensible connect/overall timeouts.
- Replace direct download calls for workflow triggers and `agentics-maintenance.yml` with the helper.
- Print clear per-file retry/failure diagnostics and exit with a concise failure summary when retries are exhausted.
- Optionally expose retry/timeout knobs via flags or env vars, and document defaults in quick-setup docs.

## Why It Won't Be That Hard
The change is tightly scoped to one script plus docs. Existing download calls are centralized and easy to swap to a helper, and no workflow compiler changes are required.

## Evidence
- `scripts/quick-setup.sh#L9` uses `set -euo pipefail`, so a transient download failure exits immediately.
- `scripts/quick-setup.sh#L188-L197` downloads each workflow using plain `curl -fsSL` with no retry/backoff.
- `scripts/quick-setup.sh#L200-L207` downloads `agentics-maintenance.yml` the same way.
- `README.md#L32-L35` and `gh-agent-workflows/README.md#L48-L51` promote the one-line quick setup path, so improving its reliability has high customer impact.

## Suggested Actions
- [ ] Add `download_with_retry()` to `scripts/quick-setup.sh` and route all setup downloads through it.
- [ ] Add focused tests or script-level validation for retry behavior and failure messaging.
- [ ] Document retry/timeouts and defaults in setup docs.

---
[What is this?](https://ela.st/github-ai-tools) | [From workflow: Trigger Product Manager Impersonator](https://github.com/elastic/ai-github-actions/actions/runs/32125034816)

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

Start with scripts/quick-setup.sh, especially lines 188-207, and inspect the existing curl download calls and set -euo pipefail behavior. Check README.md and gh-agent-workflows/README.md for the setup documentation. Done means setup downloads use retry/backoff and timeout handling, report clear failures, and focused validation covers retry behavior and exhausted retries.

Written by the indexing model from the issue text.