Integration tests

[cryptoAtwill]

Integration testing script for subnet lifecycle.

[adlrocha]

I think there are some additional code that got committed from the rebase making it harder to review.

For now, we are missing a README describing how to run the integration tests and I think we can merge this as a first iteration.

[aakoshh]

The problem with this is that it will prevent all integration tests from running, for example it will stop the ipld/resolver/tests/smoke.rs as well, which does not rely on a cluster to be set up.

Looking at the ref-fvm CI workflow, they use cargo test --exclude to prevent multiple tests from running. This only works on the crate level I think, which is why the integration tests are in a separate create within the workspace.

Other options would be to put these tests behind a feature flag and only switch them on when you want to run the end to end tests.

[adlrocha]

All of them work for me, at this point I would suggest doing whatever is easier and less cumbersome for users while keeping the door open to eventually integrating it on CI.

[aakoshh]

They might work locally, but not on CI with make test. Compare the first passing commit to the last commit:

The first runs the smoke test:

     Running unittests src/lib.rs (target/release/deps/ipc_ipld_resolver-e87f2ddc0fd49dee)

running 9 tests
test hash::tests::vector_hashing ... ok
test behaviour::content::tests::non_ephemeral_addr ... ok
test limiter::tests::basics ... ok
test provider_cache::tests::prop_providers_pruned ... ok
test provider_cache::tests::prop_providers_listed ... ok
test provider_cache::tests::prop_subnets_pinned ... ok
test provider_cache::tests::prop_subnets_pruned ... ok
test provider_record::tests::prop_tamper_proof ... ok
test provider_record::tests::prop_roundtrip ... ok

test result: ok. 9 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out; finished in 0.12s

     Running tests/smoke.rs (target/release/deps/smoke-20edbb0b97145a65)

running 1 test
test single_bootstrap_single_provider_resolve_one ... ok

test result: ok. 1 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out; finished in 3.03s

The second only runs the top section, not the bottom one.

IMO we should not let CI fall back on running the tests that it can run in the name of convenience of just not having to organise code better.

OTOH maybe it would be better to move the resolver into its own repository, since it doesn't rely on the IPC agent library at all, IIRC.

[aakoshh]

Here's a PR to move it; I don't think it's any less convenient: #153