Why you say can't and shouldn't be disputed. Still the question remains why a lot of coins claim to have low fees when there is literally no trading volume (no liquidity) at all. Transaction fees might be low, but you just can't use the coin to transfer funds as you can't get out of it anymore. That's why I think that there won't be tens of thousands of functional currencies in the future. It makes more sense for all of us to converge to a couple of them and thereby reduce all kinds of premiums, like the liquidity premium.
Consolidation down to a few coins is certainly a possibility. However, I think there is also the case to be made for a larger number of coins that, by supporting atomic swaps, be it through second-layer solutions like LN or via DEXs, provide increased resiliency against attack and centralization efforts by supporting multiple simultaneous network paths. For POS coins, the energy requirements can be mitigated by having the majority of coins use AuxPOW.
The other interesting benefit to having multiple tokens/coins is that they can provide unique utility, DGB's password manager feature being an example. Many purpose built chains that do one or two things well would be better than having a kitchen sink/jack-of-all-trades, master-of-none approach. I do agree that tokens/coins that merely tweak rewards, block-timings, etc. will cease to have value.
Everything you said is right, promising or interesting. There's nothing to argue about at this point in time as we need to see what time brings about. My theory is just this: the amount of utility or different utilities/functions provided by different tokens is essentially unlimited while liquidity is pretty much limited. I think I do see what you mean by atomic swaps and maybe thereby shifting liquidity from one network to another based on supply and demand. Otherwise I would see a couple of tokens or coins rather absorb the functionalities of other networks, thereby preventing liquidity from having to spread out through an unlimited number networks. I guess that is what you suggest will be achievable by atomic swaps. But do you think it will play out like that? This seemingly frictionless hyper network consisting of all these smaller, specific functionality networks that are connected through atomic swap technology? It is an interesting thought, but it seems so far away.