Restore bibbot arXiv detection with abstract-page metadata fallback

[Copilot]

issue_to_bibtex.py was intermittently failing to ingest valid arXiv links because the existing arxivcheck lookup returned no entries for papers that were live on arXiv. This change keeps the current lookup path, but adds a direct fallback so bibbot can still build a BibTeX entry from the arXiv abstract page.

• Lookup fallback

  • keep arxivcheck.get_arxiv_info(...) as the first lookup path
  • retry versioned arXiv IDs without the vN suffix
  • if the API-style lookup still returns nothing, fetch the arXiv abstract page and extract the publication metadata from its meta tags

• Metadata extraction

  • add a small HTML meta-tag parser in bin/utils.py
  • extract title, authors, published, abstract, canonical abstract URL, and primary category
  • preserve the existing bib entry construction logic once metadata is available

• Behavioral impact

  • avoids false “I could not find arXiv infos...” comments for valid arXiv papers
  • keeps failure handling conservative: if the fallback request itself fails, the script still degrades to the existing message path

• Example

  found, items = get_arxiv_info(arxiv_id, field='id')
  if not found and re.search(r'v\d+$', arxiv_id):
      found, items = get_arxiv_info(re.sub(r'v\d+$', '', arxiv_id), field='id')
  if not found:
      item = getArxivInfoFromPage(url, arxiv_id)
      if item is not None:
          found = True
          items = [item]

• Fixes New Papers for bibbot! #1149