External solver support in e2e tests

[devanoneth]

Description

A new function start_protocol_external_solver is created which allows e2e tests to easily setup the protocol with an external solver that uses the new solver API.

Changes

• Add a start_protocol_external_solver function
• Add a run_baseline flag to the start_protocol functions that use external solvers, which if set to true will also run a baseline solver alongside the external solver.

How to test

• CI will run, but as I'm not a contributor, forked tests can't run until merge. I've run them locally and they pass.
• Enso CI tests run and pass against this commit hash.

[squadgazzz]

LG. However, can't currently approve it since the forked node tests fail locally for some reason(even on the main branch).

[devanoneth]

LG. However, can't currently approve it since the forked node tests fail locally for some reason(even on the main branch).

Thanks! Forked node tests are working locally for me on the latest commit on my branch (just merged main in).

[MartinquaXD]

Code looks correct although a bit sketchy.
I think eventually we need to find a way to configure these CLI arguments more reasonably but this is probably a bigger endeavor.

[MartinquaXD]

This function is not used in our code base. Given that we are transitioning towards dropping the legacy API are you still using this function?
If not we should just delete it.

[devanoneth]

You're right, thanks! We don't need it anymore. Deleted in my new commit.

[devanoneth]

I think eventually we need to find a way to configure these CLI arguments more reasonably but this is probably a bigger endeavor.

I agree, there's too much string manipulation for my liking. I think something along the lines of a builder pattern to scaffold the arguments to the services would be best. Do you have some other ideas? If so, I'd be happy to do it in a follow-up PR.