Handle transient DOI lookup failures in arXiv correction workflow

[Copilot]

The arxiv_to_publications_correct workflow could fail the entire CI job when a DOI lookup returned a network error or malformed metadata. This change makes the correction script treat those cases as per-entry failures and continue processing the rest of the issue payload.

• Workflow failure mode

  • Replace uncaught requests failures with explicit handling for DOI fetch errors.
  • Add request timeouts so the job does not hang indefinitely on upstream DOI/Crossref calls.

• Metadata parsing hardening

  • Guard JSON decoding and required metadata access (author, issued) before deriving the replacement BibTeX key.
  • Skip incomplete or invalid responses with a clear log message instead of exiting the script.

• Bibliography update safety

  • Verify the referenced entry still exists and is still unpublished before mutating pint.bib.
  • Treat stale or already-updated issue content as ignorable input rather than a hard failure.

def fetch_doi_content(url, accept_header, description):
    try:
        response = requests.get(url, headers={'Accept': accept_header}, timeout=30)
        response.raise_for_status()
    except RequestException as exc:
        print(f'Ignoring {url}, failed to fetch {description}: {exc}\n\n')
        return None
    return response

[pancetta]

@copilot remove the cached files

[Copilot]

@copilot remove the cached files

Addressed in aa5587d.