Improve test coverage

[sandreim]

What needs to be done

• Implement basic functionality tests to achieve some basic/sane level of coverage.
• Fix test flakyness. Tests are using external nodes for RPC calls via subxt that can be unreachable. We should spawn a dev node using a docker image.

[AndreiEres]

@sandreim, about the test flakiness.

In tests we use external nodes for RPC calls that can be unreachable. To solve it we could spawn a dev node using a docker image.

After a few experiments with the CI team, we found that a dev node needs a couple of minutes to be ready, even with warp sync. Considering that our RPC tests are mostly smoke-like, I'd rather restart failed tests than set up a slow pipeline.

What do you think?

[sandreim]

We could spin up a small zombienet network to test this, just 1 validator and 1 parachain should do it.

[AndreiEres]

These tests are metadata sensitive and should be rewritten using dynamic API
api::tests::get_backing_groups
api::tests::get_occupied_cores
api::tests::get_scheduled_paras

[AndreiEres]

Since #445 we spawn a zombienet in the test stage instead of querying the real chain.