Issue fetching used a single per_page=100 call with no Link header follow — silent data loss once a repo exceeds 100 open autoloop-program issues.

Changes

• workflows/autoloop.md: Add parseLinkHeader() utility to extract rel="next" URL from GitHub API Link headers. Replace single-fetch with a while loop that accumulates all pages:

  let nextUrl = '...&per_page=100';
  const issues = [];
  while (nextUrl) {
      const response = await fetch(nextUrl, { headers: { ... } });
      issues.push(...await response.json());
      nextUrl = parseLinkHeader(response.headers.get('link'));
  }

• tests/: Export parseLinkHeader from the extracted JS module; add 6 unit tests covering null/empty input, single and multi-segment headers, missing rel="next", and mid-position next links.