As I explained already (see quote below), the coin age time employed to threshold the delay for signing in the variant of PoS you are using, isn't delayed by PoW computation delay. The coin age delay is a fabrication of the UXTO at that point in time. Since the attacker can construct a UXTO from his own stake and since in PoS there is no PoW computational delay impeding the attacker from rebuilding a Long Range chain attack, then the only way to prevent such an attack (i.e. the nothing-at-stake problem) in PoS is to employ checkpoints. This is is known to every expert who has studied PoS.
What is "PoW computation delay" in this context. If you are building a side chain you can assign whatever arbitrary timestamp you want to a PoW block, so there is no required delay.
The delay to compute the proof-of-work. You can't just magically pull proof-of-work out of thin air as it requires expending electricity.
Although an attacker could muck around with timestamps on his chain, he has start from some known block and he must produce a longer chain of PoW computation, which requires he consume more electricity than one the current longest chain.
These are Bitcoin101 concepts.
I added this after: Or maybe is the argument that... yes you could build a side chain that could confuse nodes up to a certain time, but that side chain will never catch up to the work added to the main chain?
So I suppose now if I am understanding the argument correctly is that.. sure you can alter the timestamps and perhaps make a confusing fork on a PoW coin, but at the end of the day it is going to be pretty much impossible to be able to have a chain that ends with the same level of computation that the main chain has.
On the other hand, using PoS, and supposing that there are no checkpoints... it would be possible to re-mine and restake all the way from the genesis block and and result in a chain that has higher trust than the main chain.
Congratulations for realizing what all of us had realized.