fix(rag): close downloaded temp file before reopening it in WordExtractor

[AlexMultiAgent]

fix(rag): close downloaded temp file before reopening it in WordExtractor

Fixes #39889

Summary

• Close the NamedTemporaryFile handle immediately after writing the remote response so the path can be reopened on Windows.
• Keep only the temp path on the instance and unlink it in close(); close() stays idempotent and tolerates a missing file.
• Update the extractor unit tests to cover path-based cleanup (unlink, idempotency, missing-file tolerance) instead of handle closing.

Root cause

WordExtractor.__init__ downloads a remote .docx into tempfile.NamedTemporaryFile() and keeps the handle open. extract() reopens self.file_path through python-docx. On Windows, an open NamedTemporaryFile cannot be reopened by another handle (PermissionError, surfaced as PackageNotFoundError), so remote .docx extraction is broken on Windows. On Linux/macOS reopening works, which is why CI did not catch it.

Validation

• Windows (Python 3.12): test_init_downloads_via_remote_fetcher now passes; WordExtractor(url).extract() returns parsed documents.
• Backend: related test_word_extractor.py cases pass on Windows; unchanged behavior on POSIX.

Checklist

• This change requires a documentation update, included: Dify Document
• I understand that this PR may be closed in case there was no previous discussion or issues. (This doesn't apply to typos!)
• I've added a test for each change that was introduced, and I tried as much as possible to make a single atomic change.
• I've updated the documentation accordingly.
• I ran make lint && make type-check (backend) and cd web && pnpm exec vp staged (frontend) to appease the lint gods

From Codex