Add a `--temp-node` flag to `forc run` for spinning up a temporary fuel-core child process for the duration of execution

Originally motivated by @nfurfaro's experience of having to repeatedly kill and restart fuel-core when testing scripts locally.

My response on slack:

> I think the way to go here would be to have a `--temp-node` or something like that that spins up an in-memory node for the duration of the `forc run` call. Rather than adding fuel-core as a dependency to `forc-client`, ideally we'd use `std::process::Command` to invoke the existing installed CLI `fuel-core` instance. I'll open an issue if we don't have one already!