[IOS-IMAP-016] Fix SwiftMail header encoding and MIME filename quoting upstream

[tabmail-kmyi]

Current status

Open. This issue tracks an upstream contribution to Cocoanetics/SwiftMail, followed by synchronization of TabMail's deviation-free SwiftMail fork.

TabMail should not carry a private fork-only remedy. The defects remain valid upstream even where TabMail has an app-side mitigation or a separately tracked app-owned surface.

The canonical technical record contains the current pin, reachability census, evidence, constraints, and related app-owned work:

• IOS-IMAP-016 canonical record

Upstream work required

• Correct SwiftMail's RFC 2047 header encoder so control-bearing values cannot be emitted literally.
• Bound encoded-words to the RFC 2047 75-octet ceiling without splitting invalid UTF-8.
• Correct all filename and content-type MIME parameter interpolation sites, with an upstream-compatible quoting/RFC 2231 policy.
• Add upstream regression tests for the full library surface, including sites not currently reached by TabMail.
• Submit the fix to Cocoanetics/SwiftMail rather than introducing a TabMail-only fork deviation.
• After upstream accepts and releases the remedy, synchronize TabMail's SwiftMail fork and update the app pin through the normal fork-sync workflow.

App-owned behavior and compatibility decisions remain tracked separately under IOS-COMPOSE-002 and IOS-COMPOSE-003; they do not replace the upstream library fix.

Completion criteria

• Upstream fix and regression tests are accepted by Cocoanetics/SwiftMail.
• A release containing the fix is available, or TabMail has an explicitly approved upstream commit to synchronize.
• TabMail's deviation-free fork is synchronized to that upstream state.
• The iOS app is re-pinned and its send/draft/render regression suites pass.
• The canonical IOS-IMAP-016 record is updated with upstream links, released commit/version, and verification evidence.
• Close this issue only after all of the above are complete; do not close it merely because an app-side mitigation exists.

Non-goals

• A permanent TabMail-only SwiftMail deviation.
• Closing the tracker as “not planned” solely because the work belongs upstream.
• Treating the app-side mitigations as proof that SwiftMail is fixed for other callers.