Random seg fault when canceling a tokio handle running a JS event loop

[CheatCod]

Intention:

Lodestone should not seg fault under any circumstances.

Why is it necessary:

Seg fault is bad and is indicative of memory access violation/corruption.

While I can't consistently reproduce this issue, the rough steps seem to be:

1. spawn multiple JS runtime within one macro executor, let $n$ be the number of js runtimes
2. cancel some runtime $a$ where $a < n - 1$, then cancel runtime $n-1$ and $n$
3. Seg faults occur in debug mode (not tested in release)

[CheatCod]

Solved with a thread-safe handle