fix(discord): truncate auto-thread and handoff thread names by UTF-16 units

[pierrenode]

What does this PR do?

_derive_auto_thread_name() and create_handoff_thread() in plugins/platforms/discord/adapter.py cap Discord thread names with a Python code-point slice (content[:80], name[:80]), but Discord validates the name field in UTF-16 code units (100 max). An emoji-heavy first message (auto-threading) or handoff name can pass the code-point cap while exceeding 100 UTF-16 units, so create_thread() is rejected with error 50035 and the thread is silently never created.

This is the sibling of rename_thread() / gateway/run.py::_sanitize_discord_thread_title(), which were already migrated to utf16_len/_prefix_within_utf16_limit — those two rename a thread after semantic titling; _derive_auto_thread_name() and create_handoff_thread() construct the name at creation time and were missed by that pass.

Fix: route both through the same utf16_len/_prefix_within_utf16_limit helpers already imported in this module. No behavior change for ASCII names.

Related Issue

No separate issue filed for this specific gap (found via sibling-site review after the recent UTF-16 truncation fixes to rename_thread() and the forum-thread-name path).

Type of Change

• 🐛 Bug fix

Changes Made

• plugins/platforms/discord/adapter.py: _derive_auto_thread_name() and create_handoff_thread() now truncate with utf16_len/_prefix_within_utf16_limit instead of code-point slices (+12/-5 lines)
• tests/gateway/test_discord_slash_commands.py: regression tests for both functions with 90-emoji names; verified RED without the fix (157/160 UTF-16 units, over the 80 budget) and GREEN with it

How to Test

uv run --frozen --extra dev python -m pytest tests/gateway/test_discord_slash_commands.py -v --override-ini="addopts="

Mutation-verified: reverting the fix locally makes both new tests fail with assert 157 <= 80 / assert 160 <= 80.

Checklist

• Contributing Guide okundu | Conventional Commits | Duplicate PR yok
• Sadece bu fix | Tests eklendi | Platform: macOS
• Docs — N/A | Cross-platform — N/A

[teknium1]

Thanks for finding the still-live auto-thread and handoff truncation gap. Current main confirms both reported code-point slices at plugins/platforms/discord/adapter.py:5332-5335 and :5495-5521, and the proposed use of the existing UTF-16 helpers matches the current semantic-rename implementation.

Problems

• The same creation-time class remains in native /thread: _create_thread() only strips name at plugins/platforms/discord/adapter.py:5256 before supplying it to both create_thread() calls at :5279-5299.
• Forum thread names still use first_line[:100] at plugins/platforms/discord/adapter.py:7787-7794; those names flow to forum creation at :2119-2127, :2188-2199, and :7975-8033.

Suggested changes

• Extend this UTF-16 pass to those sibling creation paths using _prefix_within_utf16_limit() and each path's intended budget.
• Add emoji-heavy assertions for /thread and forum creation, alongside the two good regressions already proposed here.

Automated hermes-sweeper review.

[pierrenode]

Rebased onto current `upstream/main` — production code (`plugins/platforms/discord/adapter.py`) auto-merged cleanly, fix intact. The test file conflict was upstream's own pruning of several pre-existing `_auto_create_thread` tests unrelated to this fix — kept upstream's pruning, restored only this PR's own new UTF-16 regression test.

Mutation-verified: reverting `_derive_auto_thread_name`'s truncation to code-point slicing breaks the new test (157 UTF-16 units instead of ≤80). Full `tests/gateway/test_discord_slash_commands.py` (19 tests) passes. Ruff clean. Fresh competitor search: #60196 (forum thread names, `_derive_forum_thread_name`) touches a different, sibling function — no overlap with this PR's `_derive_auto_thread_name`/`create_handoff_thread`.

Squashed to a single commit on top of current `upstream/main`.