Itetst: Cleanup ExpectedInvocactions

[arajasek]

The current expectations lead to a lot of bloat in the tests. It may not be too easy to avoid this, since they need to be fairly flexible, but there's likely some helpers that can be created to simplify this.

[LesnyRumcajs]

@arajasek @ZenGround0
I was thinking roughly of an API like this, kind of inspired by what clap provides:

ExpectInvocation::new()
    .to(CronActor)
    .method(CronMethod::EpochTick)
    .sub_invoc(vec![
        ExpectInvocation::new()
            .to(StoragePowerActor)
            .method(PowerMethod::onEpochTickEnd),
        ExpectInvocation::new()
            .to(RewardActor)
            .method(RewardMethod::UpdateNetworkKPI)
        ])
    .assert_match(v);

that would replace

    ExpectInvocation {
        to: *CRON_ACTOR_ADDR,
        method: CronMethod::EpochTick as u64,
        subinvocs: Some(vec![
            ExpectInvocation {
                to: *STORAGE_POWER_ACTOR_ADDR,
                method: PowerMethod::OnEpochTickEnd as u64,
                ..Default::default()
            },
            ExpectInvocation {
                to: *REWARD_ACTOR_ADDR,
                method: RewardMethod::UpdateNetworkKPI as u64,
                ..Default::default()
            },
        ]),
        ..Default::default()
    }
    .matches(v.take_invocations().last().unwrap());

In essence, it gets rid of manual casts in every method, having to call Default::default and whatever could be abstracted away so even a non-Rust dev could understand what is happening.

We could get even fancier with a limited amount of macros if we expect hundreds of such tests but IMO this alone would make the ITGs codebase a bit cleaner and easier to read.

[anorth]

integration_tests/src/expects.rs added a bunch of helper methods that in practise reduce a lot of the noise in tests. The pattern is limited by not handling sub-invocations, though.

[rjan90]

This issue is related and might supersede this one: #1548 ?