Restore and update mempool tests

[conradoplg]

Motivation

See #2959

Specifications

Designs

Solution

Restore and update the tests. In most cases, instead of working with a transaction count, a cost limit was derived by summing the costs of some of the transactions that will be inserted.

Closes #2959

Review

I just focused on restoring the tests and still want to think a bit more about if these tests still test everything that we want or if something is missing. But since we want to get this done soon I'm opening the PR as is. Feel free to suggest additions, otherwise I can also amend the tests in a later PR.

Anyone can review, but @teor2345 and/or @dconnolly might be particularly interested.

This seems to be one of the last pending issues for the beta so it's best to review it soon.

Reviewer Checklist

• All deleted tests were restored
• Code implements Specs and Designs
• Tests for Expected Behaviour
• Tests for Errors

Follow Up Work

[teor2345]

Looks good to me, thanks for making this happen!

[teor2345]

It looks like one of the tests sometimes fails:

---- components::mempool::storage::tests::vectors::mempool_storage_basic stdout ----

The application panicked (crashed).
Message:  assertion failed: !some_rejected_transactions.is_empty()
Location: zebrad/src/components/mempool/storage/tests/vectors.rs:104

https://github.com/ZcashFoundation/zebra/runs/4029234056?check_suite_focus=true#step:10:1124

This intermittent test failure is a blocker for merging this PR.

[teor2345]

The same failure also happened in the coverage build, along with:

 ---- components::mempool::tests::vector::mempool_queue stdout ----
Error: 
   0: mempool storage has a cached chain rejection for any transaction with the same effects
   1: transaction evicted from the mempool due to ZIP-401 denial of service limits

Location:
   /rustc/e269e6bf47f40c9046cd44ab787881d700099252/library/core/src/result.rs:1915

[teor2345]

If the randomised tests are quick, we might want to run them multiple times, to make sure we catch any intermittent failures.

[conradoplg]

If the randomised tests are quick, we might want to run them multiple times, to make sure we catch any intermittent failures.

Good idea! I did that and fixed the tests in ca21735

[dconnolly]

I'm fine with these retries for now 👍

teor approved then requested changes for failing tests that are now fixed