nowhere it is written that the bullet order must be respected so it is not in the spec IMO

I am sorry, but that sounds like a rather poor argument. I am pretty sure you can find other parts in EE specs where you would happily assume ordering based on a written list. For instance Container Lifecycle Events also list certain events under Application lifecycle events, that are fired once. And I think we can all safely "assume" that BeforeBeanDiscovery somehow precedes {{BeforeShutdown }}even though the spec doesn't say that the bullet order must be respected, right?

Also, if you twist it around, then having it as the test presumes allows you to introspect and change things around in PIP and PIT and then based on the current state decide whether you want to veto the bean or not inside PBA. This to me sounds a lot more sensible than firing PAT with veto option followed up by PBA with the exact same option when nothing really changes in between.

Not directly linked to this issue but since you mention it: is 2.0.1.Final anything in terms of spec (I can see what it is in terms of artifact release but not spec)?

Not sure I understand what you ask about? There are multiple releases of TCKs for each CDI spec versions. Typically containing new exclusions, test corrections and so on. Take a look here - the 2.0.x TCK version is linked to CDI 2.0. Latest for that version is actually 2.0.7.SP1.