2) Enforcing a hybrid chain (with alternating PoW/PoS blocks) is IMO a bad idea as the protocol is set to 1-minute PoW and 10-minute PoS target. With my rules hybrid chain is the optimal way when trying to do 51% attack (as you can reduce the "51%" PoW hashing power needed by at most 50% if you have 100% active weight). Still, it would not be expensive enough to attempt, anyway - that's my justification of lowering PoS trust to the level of PoW (I'd lower it even more if only it didn't cause another sort of problems, which it would). It's all about finding the right balance, anyway.
It's not a "good" or "bad" way. It's how the things should be.
Preferred, but not
enforced chain. For example, you can use own function of blocks share here, to maximize trust score for 10:1 chain and minimize it for another candidates.
Variable ROI isn't sufficient to prevent a malicious entity wanting to break the network entirely, anyway.
It's only a part of solution. It makes malicious activity to be a less danger for a network by increasing the share of coins participating in the network protection.
Good example of simpler != better statement. It will help you for one threat, but opens another hole. Actually, such fix is less secure than calculate block trust using an original algorithm. It can be forked without a significant part of stake or hashpower by running a parallel chain at lower PoS & PoW difficulties. Because it makes no difference between coindays consumed or hashpower wasted. One CPU is able to beat the entire network.