fix(weixin): handle cross-event-loop session reuse in send_weixin_direct

[eipiem1]

Problem

When cron jobs deliver results to Weixin, send_weixin_direct() attempts to reuse the cached live adapter's aiohttp session (created on the gateway's event loop). However, standalone cron delivery runs via asyncio.run() on a different event loop.

aiohttp's TimerContext.__enter__() calls asyncio.current_task(loop=gateway_loop), which returns None on the standalone loop, raising:

RuntimeError: Timeout context manager should be used inside a task

This causes all cron deliveries to the Weixin (WeChat) platform to fail silently.

Fix

Wrap the live-adapter reuse path in try/except RuntimeError so it falls through to creating a fresh aiohttp.ClientSession on the current event loop. The fresh session path was already implemented — it just wasn't reachable when a cached live adapter was present.

Impact

• ✅ Cron → Weixin deliveries now work correctly
• ✅ Gateway → Weixin deliveries (same event loop) continue to work unchanged
• ✅ No performance regression — live adapter is still preferred when on the same loop
• ✅ Minimal change: ~12 lines added

[alt-glitch]

Likely duplicate of #14481, #12810, #15911 — same cross-event-loop aiohttp session reuse in send_weixin_direct causing RuntimeError on cron delivery. Multiple open fix PRs already exist.

[alt-glitch]

Likely duplicate of #14481, #12810, #15911 — same cross-event-loop aiohttp session reuse issue.

[eipiem1]

Thanks for the heads up @alt-glitch! Closing in favor of the existing PRs (#14481, #12810, #15911). The fix was already known — appreciate the pointer.