I don't understand the point about the "quadratic validation" problem, but the first two issues of undermining consensus seem sufficient in order to cause the proposal to be totally and completely unacceptable.
Validation time is quadratic. This means that somebody could construct a transaction (at 2 MB) that would take too long to validate (over 10 minutes). BIP109 does not solve this, but adds a limit to the size of transactions. Segwit on the other hand scales down the validation time making it linear.
Does Core intend to go ahead with the proposal reached at the roundtable consensus? (segwit ASAP + 2MB HF a year later)
Again, you didn't understand the roundtable at all. The people who were present there (some Core developers) have to present a HF proposal and code between April and July. This does imply that this proposal will be implemented/accepted by either developers/miners/community (i.e. no guarantee). You first have to wait for the actual proposal. How many times do I have to tell you this?
You're either being naive or disingenuous.