The correct order - as far as I remember it has been discussed originally - is

So, I don't know what you discussed and where/when, but current spec wording is pretty straightforward.

If you look at Bean discovery section, it defines the ordering for both, class based bean and producer based bean. This exactly corresponds to what the test asserts.

Note also, that this TCK test is not part of the official CDI-2.0 JSR but added at a random later point.

The "random later point" can be divined from GH commit history - seems this got in with 2.0.1.Final even though the PR linked here was actually closed.

Spec lists the needed events but it is not written the order must be this oneSpec lists the needed events but it is not written the order must be this one

That's just bending the wording. Is there an actual reason why this ordering is bad? I can't see why you couldn't veto things in PBA with the current ordering, it isn't too late.