fix(tui_gateway): offload remote-media attach handlers to the RPC pool

[vondelomlo]

What does this PR do?

Stops the gateway from freezing while a remote PDF or image upload is processed.

dispatch() only offloads methods listed in _LONG_HANDLERS to the thread
pool; everything else runs inline on the single reader loop, which reads the
next RPC only after the current handler returns. The remote-media-relay
feature (16786f3) added two genuinely long-running handlers that were left
off that list:

• pdf.attach shells out to pdftoppm with a 120s timeout.
• image.attach_bytes base64-decodes and writes multi-MB uploads to disk.

While either runs inline, inbound RPCs — notably session.interrupt and
approval.respond — sit unread in the pipe/socket, so the UI appears hung for
the full render/decode duration and the user cannot interrupt the turn or
answer an approval prompt. The old image.attach only resolves a local path,
so it stays correctly inline.

Fix: register both new handlers in _LONG_HANDLERS so dispatch() routes them
to the pool and frees the reader loop immediately — the exact pattern already
used by slash.exec, shell.exec, and session.compress.

Related Issue

N/A

Type of Change

• 🐛 Bug fix (non-breaking change that fixes an issue)

Changes Made

• tui_gateway/server.py: add pdf.attach and image.attach_bytes to the _LONG_HANDLERS frozenset; refresh the explaining comment to cover the media-relay handlers.
• tests/tui_gateway/test_protocol.py: add a parametrized test proving each media handler is offloaded and a concurrent fast handler completes in <0.5s instead of blocking behind it.
• scripts/release.py: register the contributor email in AUTHOR_MAP.

How to Test

1. scripts/run_tests.sh tests/tui_gateway/test_protocol.py — full file green.
2. Targeted: pytest tests/tui_gateway/test_protocol.py -k "media or dispatch" -q — 16 pass, including test_dispatch_media_attach_does_not_block_fast_handler[pdf.attach] and [image.attach_bytes].
3. Before the fix the new test blocks ~5s on the fast handler and fails the <0.5s assertion; after the fix it passes.

Checklist

Code

• I've read the Contributing Guide
• My commit messages follow Conventional Commits
• I searched for existing PRs to make sure this isn't a duplicate
• My PR contains only changes related to this fix
• I've run the relevant tests and they pass
• I've added tests for my changes
• I've tested on my platform: macOS 15.5

Documentation & Housekeeping

• I've updated relevant documentation (docstrings/comment) — or N/A
• I've updated cli-config.yaml.example if I added/changed config keys — N/A
• I've updated CONTRIBUTING.md or AGENTS.md if I changed architecture or workflows — N/A
• I've considered cross-platform impact — N/A (pure dispatch routing, no platform-specific code)
• I've updated tool descriptions/schemas if I changed tool behavior — N/A

[teknium1]

Thanks for the focused fix. The premise still holds on current main: tui_gateway/server.py:1271 invokes every method absent from _LONG_HANDLERS inline, while image.attach_bytes performs decoding/disk writes (tui_gateway/server.py:9601-9619) and pdf.attach runs pdftoppm with a 120-second timeout (tui_gateway/server.py:9723-9728). Adding these two names follows the existing pool-routing contract.

The related merged PR #53895 covered the separate complete.path and complete.slash handlers, not these media methods.

Automated hermes-sweeper review.