FFI - Tracking leaked SVM errors

[YaronWittenstein]

In a very similar way to the #401 issue - this issue asks to add tracking for leaking errors.

When the Wasm code running panics, the Wasm runtime (Wasmer) catches it and finishes its execution gracefully, and the panic isn't propagated back to the Rust code calling Wasmer. Thus, an error returned from the FFI layer is usually a bad sign since it implies that SVM has panicked.

Here is where the code builds an error to be returned: https://github.com/spacemeshos/svm/blob/master/crates/runtime-ffi/src/api.rs#L522

And this is where the leaking takes place: https://github.com/spacemeshos/svm/blob/76be2e2e7f701be0cb0de6cb226b5166ce730fda/crates/runtime-ffi/src/result.rs#L50

[neysofu]

Closing - not relevant; much like #401.