Greptile Summary

This PR adds a small, targeted heuristic (tryRecoverLatin1AsUtf8) to fix garbled CJK filenames when the Feishu API returns a plain filename="…" Content-Disposition header instead of the RFC 5987 filename*=UTF-8''… form. The HTTP client decodes the raw UTF-8 bytes as Latin-1, and the helper re-encodes those Latin-1 chars back to bytes and re-decodes as UTF-8, falling back to the original string if UTF-8 decoding produces replacement characters.

• The \uFFFD guard is the correct way to detect a failed UTF-8 decode in Node.js and prevents false positives for most real Latin-1 content.
• The empty catch {} is appropriate since the function always falls back to the original string on any error.
• The change is narrowly scoped to the plainMatch path; the existing filename*=UTF-8'' (RFC 5987) path is unaffected.
• One inherent limitation: a Latin-1 filename whose byte sequence happens to be valid UTF-8 (e.g. two Western-European characters forming a 2-byte UTF-8 sequence) would be silently rewritten. This is an unavoidable trade-off with any header-recovery heuristic and is a very unlikely scenario in Feishu's environment.

Confidence Score: 4/5

• Safe to merge; the fix is minimal, well-guarded, and falls back gracefully.
• The change is a single, self-contained helper with a correct fallback mechanism. The \uFFFD guard and try/catch block protect against incorrect recovery. The only deduction is the inherent heuristic ambiguity (valid Latin-1 that forms valid UTF-8), which is an accepted trade-off and very unlikely in this context.
• No files require special attention.

_{Last reviewed commit: cd566af}

ProjectClownfish could not safely update this branch, so it opened a narrow replacement PR instead.

Replacement PR: #72388
Source PR: #48578
Contributor credit is preserved in the replacement PR body and changelog plan.