In the case of 4, well, its just a disaster. Blocks can be replaced all the way back to the last checkpoint potentially and all transactions from that point could be destroyed.
You aren't capturing the realistic effect of a 51% attack. Society is not going to allow all historic blocks to be replaced. Politically impossible. Instead the viable attack is insidiously changing the protocol going forward, such as censoring transactions which violate some government edict, e.g. forcing every transaction to be stamped with its government KYC identification number (a la 666).
In any event its disastrous due to the agreement being based on resource expenditure and not the data set.
The 51% attack is allowed to go unchecked = centralized
Reliance on developer inserted checkpoints = centralized
Government intervention = centralized
Your design will also be FUBAR under a 51% level attack on the consensus.
Also there is no design yet for crypto currency that is valid which doesn't consume a resource. I summarized upthread that Proof-of-Stake is more flawed than Proof-of-work, because the resources is consumed only once and is not ongoing consumption.
You will not be able to find a solution that doesn't consume a resource. I nearly 99.9% guarantee that. These attempts to use propagation (e.g. VanillaCoin) as a consensus are doomed.
Rather I am aiming (if I am satisfied with the tradeoffs I enumerate in my design as I continue to reanalyze it) to improve upon the way the resource consumption is organized, so the economics of mining and blocks are improved for the properties we desire (e.g. see my point about the ability to fork away from the dumb masses). Also to enable instant transactions.
As will anyone's, except with regard to eMunie consensus at least, the absolute critical difference is that the data BEFORE the attack is still valid and can be trusted as a true record.
Bitcoin can not provide that assurance as an attacker can rewrite history which is my major gripe with hash POW.