Should thread.spawn be fallible?

[fitzgen]

Should thread.spawn be fallible, and return -1 on failure (similar to memory.grow)? It represents large allocations by the engine, such as the stack for the thread, and these allocations might fail. Do we want that failure to trap and kill the Wasm instance and everything else in the store, or do we want to return an error code and let Wasm recover?

[yamt]

+1 to make it return -1 on failure.
from POV of apps, it's often recoverable.