Summary

Extends scripts/buildtools/check_tests.py to customize tests to ignore via --ignored-tests. The motivation is to acquiesce check_tests.py warnings about E2E expect tests run multiple times.

In another branch, I saw warnings in https://app.circleci.com/pipelines/github/algorand/go-algorand/8787/workflows/a566ab26-42ad-48de-940d-08d504ce1c15/jobs/158801?invite=true#step-103-7. Reproduced below for convenience:

=========== RAN MULTIPLE TIMES ===================
The above 4 tests ran multiple times:
github.com/algorand/go-algorand/test/e2e-go/cli/algod/expect TestAlgodWithExpect -- ran 10 times. (Can probably be fixed by adding "partitiontest.PartitionTest()")
github.com/algorand/go-algorand/test/e2e-go/cli/algoh/expect TestAlgohWithExpect -- ran 10 times. (Can probably be fixed by adding "partitiontest.PartitionTest()")
github.com/algorand/go-algorand/test/e2e-go/cli/goal/expect TestGoalWithExpect -- ran 10 times. (Can probably be fixed by adding "partitiontest.PartitionTest()")
github.com/algorand/go-algorand/test/e2e-go/cli/tealdbg/expect TestTealdbgWithExpect -- ran 10 times. (Can probably be fixed by adding "partitiontest.PartitionTest()")
The above 4 tests ran multiple times:
==================================================

Upon inspection, I realized these tests are parent E2E test runners for expect tests. They're invoked multiple times by-design and the advice to add partitioning does not apply.

As a band aid, the PR exposes a way to ignore these tests. See the PR's builds for proof that the warnings disappear.

• Treat the PR optionally. I didn't see a way to prevent the tests from reporting via go test flags. I can envision managing the ignore list as worse than the status quo.
• I realize the verification step happens for all test types. The configuration relies on test name uniqueness.
• For the PR's scope, I'm gauging if the band aid is deemed a net positive. If it isn't, I'd prefer to close and move on.

Test Plan

Run CI.