To your point: the way to fix these problems is to make sure that everyone, irrespective of their ability to make money selling their labor, is able to consume a fair amount. Hence the need for a basic income for all: the higher the better. The quicker we break from the moralistic perspective that everyone should "work" in a world where not everyone can, the quicker we can create a world that is about "the pursuit if happiness" again.
The fact is that we don't live in the long term so although truck driving and coal mining are shitty jobs (and absolutely should be automated or eliminated) we have to take care of the current practitioners (both financially and finding a way for them to preserve their dignity) as their employment goes away.
Separately I don't think we'll need all that many robot designers either.
If there is shortage of skilled workers to do those jobs then they should train them too. The lack of long term job security has lead to the end of loyalty which means private companies have no incentive to train people which means the unemployed have to take on debt to study courses which aren't necessarily giving the skills they need. That and companies need to start sponsoring and training and accept that some will end up at a competitor.
While the article doesn't explicitly mention it, I'd assume that they're looking at productivity in real, not nominal figures, precisely to disentangle the effects of inflation.
Meaningful comparison across countries requires purchase power parity (PPP) exchange rates, rather than market exchange rates. However, this article doesn't really talk about other countries at all, so I don't understand why you bring up PPP here. If fewer hours of labour are required to buy dinner, that's just real wages, nothing to do with PPP...
And how does lower labour force participation enter? (If we assume that less productive workers drop out first, that should increase productivity, btw - which we don't see...)
I think a lot of the productivity mystery lies is due to artifacts in the shitty tools we have. For example, GDP is like the Dow Jones indices: designed to be easy to calculate with pencil and paper, plagued by necessary changes in their own baselines over time (mostly: what is counted and what is not), not terribly reflective of what we really want to know, and yet it would be stupid to discard them because we'd lose any time series. Everybody knows they are broken but nobody has any idea yet what would be better.
It's not like I have a great idea of what we should be using instead ettither. But I fell it's foolish to say "we have strong qualitative signals but since we can't find them in our model they are bogus". Especially since we all know our current theories are woefully incomplete and perhaps even inaccurate.
Those who don't see the productivity advance aren't idiots: those who, like me, believe in those signals justifiably have a large hurdle to leap.
Clearly there is geographical variation! But low inflation is considered a trap by many mainstream economists (I disagree but appreciate their reasoning).
Just look at the number of hours of typical labor required to buy many goods (TVs, meals, etc) and compare it to your parents' time and you can see the low inflation in action.
Yes, GDP$ per unit hour has not gone up much, but the flip side is that the economic value of the output has gone down. If you look in terms of PPP, workers are doing better: fewer labor hours go into buying dinner.
The corollary is that fewer hours are required to produce PPP equivalent product (e.g. a car) so fewer workers are needed (since the $ are not going up at the same rate it looks like productivity is not). That's why the LFPR is sinking.
A victorian era "smokestack" model isn't useful when looking at this world. This is why automation really is likely to cut employment this time (in fact it already is) and that's the real economic problem to be worrying about. Which some are (e.g. the small UBI experiments under way) but it's still not mainstream.