More nits

Fix nits

The comment is still accurate. The behavior did not change, as tests demonstrate.

Address some nits

Let's wait until tomorow for this, just in case.

Now of course for the most strenuous activation tests (involving consensus features) it's important to actually step through the activation block by block, see e.g. D5179 and D5130. That lets you check very carefully how certain transactions in mempool before the upgrade are handled during the upgrade, and vice versa -- if the upgrade is rewinded, the mempool needs to be sane still. But as this is nonconsensus it's not really important.

Enforce phonon activation time so that test pass also after the upgrade.

Remove change to miner_tests, as it is covered by the intergration test and messes with global state.

What @markblundeberg said.

Update to use a whitelist only and bump to 5% after discussions with miners.

The builds are clean.

The code looks correct, but the test plan is inadequate.

Patch doesn't touch the seeds.

You need to update the cmake build.

This is much better, but is missing cases. Thankfully, it is now possible to see which cases are tested an which are not by now.

I get what you want to do with this whole stack, and this is not a bad idea per, but you need to rework the approach so that flagset getting out of sync is less of a problem.

Wait, what did happen here?

Some details can be improved, but it is good overall.

If the files are not touched, it doesn't make sense to update the copyright.

This is a bit of a concern, because now a dev adding a new test suite needs to update this to make sure it runs. Failing to do so will just result in not running the test suite, silently. You need to figure this out.