fuzz-lightyear should support running phet-io wrappers

[zepumph]

This is blocking migration work because it is untenable to use only CT to report the problems we are encountering.

[zepumph]

• [x] I also got the go ahead to remove broken and old code from test-server related to local builds. This hasn't worked in a long time and likely wouldn't be the starting point for this feature in the future.

[zepumph]

I am so happy now that we removed the build related logic. Now we get much better highlighting when we succeed.

I would like to ask @samreid about what wrappers we might want to support here. Right now it is just migration and studio.

[zepumph]

• [x] Support state wrapper, confirm that the "loaded" metric makes sense and works well.
• [x] Rename testDuration in its current form to "timeout"
• [x] Add a new implementation for testDuration that only starts the timer once we are loaded.

[zepumph]

I would like to do a couple more things here:

• [x] PSA on slack
• [x] Bring to dev meeting so people are aware
• [x] Give this a new name. Right now it is called test-server. It isn't a server anymore, and I think we can do better:
  • super-fuzzer
  • local-fuzzer
  • fuzz-master
  • fuzzer-extraordinare
  • fuzz-lightyear

[phet-dev]

Reopening because there is a TODO marked for this issue.

[samreid]

Here is the existing TODO:

              // TODO: some older sims don't support ?wrapperContinuousTest and so we are loaded if just the first goes. https://github.com/phetsims/aqua/issues/190
              // To also support the basic 'load' event to match the language of the sims
              window.parent && window.parent.postMessage( JSON.stringify( {
                type: 'load',
                url: window.location.href
              } ), '*' );

I'm not really sure how it should be addressed.

[zepumph]

I think if we MR'd support for continuous-test-wrapper-load for the oldest of our migration support wrappers then we would be able to solve this todo (getting rid of this postMessage) AND potentially fix the issue we have a better solution for https://github.com/phetsims/phet-io-wrappers/issues/507. At the very least, perhaps we should rely on wrapperContinuousTest as a parameter for the migration wrapper to squeeze out load events from both studio versions. That way we never ask for an upstream state prematurely.

[zepumph]

Discussed with @jbphet @samreid. While we are at it, we should make sure to rewrite the migrate.js code to always listen to the wrapperContinuousTest logic to achieve the "have both studios loaded". Perhaps we should have an MR for it and create a new issue.

[zepumph]

This will be done in https://github.com/phetsims/phet-io-wrappers/issues/507, closing.