It’s important to note that the theoretical attack in this post hinges upon counting equivocations which the beacon chain spec does not do. So “creating zillions” of fake block proposals is not made worse by the above.
[Apologies for hijacking the thread. Such an attack if double counting equivocations in LMD GHOST is a very valid and viable threat]
All of these reported issues were responded to. The particular issue you noted here was with respect to DoS attacks to which the DoS protection measures found in the p2p spec were referenced. The approximate number of DoS blocks you could get onto the network is approximately a function of the count of honest nodes on the network (assuming you can deliver a unique block to each honest node before it has seen another block forwarded from another honest node), so “a zillion” is not the bound on the DoS attack.
Also noted, is that with simply an equivocation block DoS attack, subsequent honest proposals would quickly coalesce on a single chain (barring some more sophisticated balancing attack).
There is an ongoing bug-bounty program if that interests you – Ethereum Bug Bounty Program | ethereum.org
Additionally, testnets or private networks are a good place to attempt attacks before submission to the bug bounty program.