fix: unpark correct parker on resize abort

[Qian-Cheng-nju]

Hi, thanks for papaya!

When a blocking resize is aborted, help_copy_blocking (raw/mod.rs:2073) unparks table.state().parker, but threads waiting for the resize are parked on next.state().parker. The unpark goes to the wrong parker instance and parked threads are never woken, causing a deadlock.

The fix changes table.state() to next.state() on the unpark call so it targets the parker where threads are actually waiting.

[Qian-Cheng-nju]

Hi @ibraheemdev, just checking if you had a chance to look at this — happy to address any concerns!

[ibraheemdev]

This seems correct, did you run into a deadlock caused by this? It's unfortunate that we don't really have any test coverage for this code path because it's quite difficult to trigger.

[Qian-Cheng-nju]

We didn't hit it at runtime — we were reading through the resize code and noticed that help_copy_blocking parks on next.state().parker but the abort path unparks table.state().parker. Since they're different Parker instances, any thread parked during an abort would never get woken. It's definitely a rare path, but seemed worth a fix since it's a one-liner.