Check out "Debt: The First 5000 Years", which puts forth decent reasoning that the "barter -> currency -> lending" path is not accurate but rather a bad post-hoc analysis of what we think might have happened that is based off of "what if modern society didn't have currency", as opposed to "how did real societies before currency look".
From that basis, it suggests a likelier path of "local lending -> currency", i.e. that a straight up barter system never existed in ancient societies, and instead a notion of "who owes whom" was publicly known in small tribes and while those with the power to enact violence often had the ability to demand things somewhat disproportionately, most people did a good job of assessing what stuff to give, what people to give it to, and when to do so, as a moderate sized socialable semi-cooperative set of families with common interests might be capable of just... knowing.
When tribes began trading things to each other at scales too large or untrustworthy to keep track of by memory, the simple innovation of writing it down in some form would often occur, which is the birth of currency, long after the notion of indebtedness, and other innovations of abstraction and regulation and such can build on top of that over time to create the modern concept.
This doesn't really speak to your major point that a modern currency must support a large number of transactions/minute with low or no fees, which I agree with and agree Bitcoin seems (for now) ill-suited to handle; I've seen suggestions (on a paper linked on marginal revolution I think) that Bitcoin makes more sense as an asset to hold like gold or stock than it does as a currency, because its transaction fees, timing, and (for the moment) growth characteristics strongly incentivize not spending it.
But what I was saying was that currency is not important because "without it we'd all be bartering", rather, it's important because "without it we would need to trust and know everyone we deal with every day well enough to track all our mutual indebtedness mentally", Which is plainly impossible in today's world!
Currency needs to have few reasons to hold onto it rather than use it, and needs to be able to be exchanged quickly between any pair of actors regardless of external state. Cash is great for that, credit cards are getting there (me->business is good, business->business and business->me and me->you all much less so), and Bitcoin is a long, long way off.
To tag some more weight into this very long post: my personal bet is that networks of micro-applicable cryptocurrencies (say, "Toronto Downtown Ethereum") in a network-system like Cosmo that settles debts across multiple blockchains quickly and efficiently will approach the use-case of cash, with the properties of local, highly scalable, and low-fee transactions. We'll all be using baskets of currencies all the time, basically, but with that fact abstracted for most people down to card or phone tap actions. Some professionals will still care which currencies, and how they operate, but they'll be a bit like forex traders are today.
To make this prediction interesting, I give it a 15% chance at being reality somewhere within 3 years, a 50% chance in 5 years, 80% in 10 years, and 20% chance it takes longer than that or never happens because new technology/events make us skip/avoid it. (these averages obviously must accumulate the earlier cases within them where needed.)