Common code unit tests are not run in PhET-iO brand

[zepumph]

Looking through listContinuousTests.js, I don't see a way that common code is being tested in phet-io brand. This is because repos like tandem and axon are not in perennial/data/phet-io, as they aren't runnable phet-io repos. For right now, I will just manually add them to make sure they get unit tested in phet-io brand.

[zepumph]

After the above commit, I'm pretty sure that axon is going to start failing on CT, as that was what I found to be the case when beginning to work on phetsims/axon#316, but it is better for it to be load failures instead of quiet ones.

Tagging @jonathanolson to make sure that it is alright to use assert in listContinuousTests.js as I did above.

Tagging @samreid and @chrisklus to note that we have failing axon unit tests in phet-io brand (another issue coming soon). And also to note that this is likely not complete, we my want to brainstorm more complete solution.

[zepumph]

Also, I wonder if this would be solved by adding phet-io unit tests to pre-commit hooks.

[zepumph]

Here are the tests axon was testing up until this point:

[samreid]

Would it be reasonable to say "all tests run in ?brand=phet-io" and we endeavor to make the phet brand unit tests a subset of phet-io brand unit tests?

[zepumph]

Yes, as long as we don't just run them in PhET-iO brand. Perhaps they could all be run in both on CT? I don't think that would require much work.

I ran the following, and found potentially simple instrumentation errors to fix in scenery, scenery-phet, and sun. Everything else passed:

http://localhost:8080/area-model-common/area-model-common-tests.html?ea&brand=phet-io
http://localhost:8080/area-model-common/area-model-common-tests.html?ea&brand=phet-io
http://localhost:8080/axon/axon-tests.html?ea&brand=phet-io
http://localhost:8080/balloons-and-static-electricity/balloons-and-static-electricity-tests.html?ea&brand=phet-io
http://localhost:8080/circuit-construction-kit-common/circuit-construction-kit-common-tests.html?ea&brand=phet-io
http://localhost:8080/dot/dot-tests.html?ea&brand=phet-io
http://localhost:8080/faradays-law/faradays-law-tests.html?ea&brand=phet-io
http://localhost:8080/fractions-common/fractions-common-tests.html?ea&brand=phet-io
http://localhost:8080/griddle/griddle-tests.html?ea&brand=phet-io
http://localhost:8080/interaction-dashboard/interaction-dashboard-tests.html?ea&brand=phet-io
http://localhost:8080/joist/joist-tests.html?ea&brand=phet-io
http://localhost:8080/kite/kite-tests.html?ea&brand=phet-io
http://localhost:8080/phet-core/phet-core-tests.html?ea&brand=phet-io
http://localhost:8080/phet-io/phet-io-tests.html?ea&brand=phet-io
http://localhost:8080/phetcommon/phetcommon-tests.html?ea&brand=phet-io
http://localhost:8080/scenery/scenery-tests.html?ea&brand=phet-io
http://localhost:8080/scenery-phet/scenery-phet-tests.html?ea&brand=phet-io
http://localhost:8080/studio/studio-tests.html?ea&brand=phet-io
http://localhost:8080/sun/sun-tests.html?ea&brand=phet-io
http://localhost:8080/tandem/tandem-tests.html?ea&brand=phet-io
http://localhost:8080/twixt/twixt-tests.html?ea&brand=phet-io
http://localhost:8080/utterance-queue/utterance-queue-tests.html?ea&brand=phet-io

[zepumph]

I was a little on the fence thinking about all of the above sims that don't support PhET-iO brand, but especially after the above commit, I think that it is still the way to go.

The above work now makes scenery run successfully in scenery. I would say it was 90% busy work, but it still felt like important clean up. The other 10% were actual PhET-iO tests that were broken and hadn't been run anytime recently. Wahoo!!

Next steps for this issue:

• Fix scenery-phet and sun unit tests
• Have CT run unit tests in phet-io brand.
• I think we should run unit tests in phet-io brand as the pre-commit hook over in What brands should we run unit tests for in the precommit hook? #174. After this issue, it will be straightforward to make it happen!

[zepumph]

@samreid please review in correlation with #174. How do we feel about the assumption that all unit tests support being run in brand=phet-io?

[samreid]

Running all unit tests in both brand=phet and brand=phet-io sounds appropriate and in line with the long term vision of our project. For repos that have no special phet-io features, they will run redundant tests, but I believe that is OK because most unit tests are fast to run.

If and only if we decide that is taking too long, then I'd recommend pursuing a strategy where we only (or predominantly) run tests in phet-io brand, and make sure the phet brand tests are run as a subset. For the rare cases where we need to test something in phet-brand, that could be listed as a special case.

I skimmed the commits and didn't have further recommendations. Thanks @zepumph!