But there's reasons why they weren't maintained: it was a lot of work, and people didn't play them (who came first, the chicken or the egg

). Too much work for too little gain.
Well yes that's the point. HOI3's development process was vastly different than HOI4's, so we're not talking about analogous evidence here.
Nothing wrong with that. If you have a product that's costly and time-consuming to make, and your consumers/customers do not care for it, that's a pretty strong argument to discontinue that product.
Read the initial post. 1939 is a bad data point.
He basically says that the earlier start dates (plural!) were the most popular in HOI3, so they were the ones incorporated in HOI4. Presumably, 194x start dates were even less popular than the 1939 start date. In his second quote he confirms that adding further start dates doesn't make any sense for Paradox. You know, based on their knowledge of player usage and costs, which you, as you stated yourself, do not possess.
Additionally, we can gather that the 1939 start date in HOI4 is less popular than in HOI3.
To put it in other words: HOI3 showed that the 194x start dates weren't justified by player usage, so they were left out in the next iteration. HOI4 showed that the 1939 start date isn't justified by player usage (and will probably be left out in HOI5, in days to come).
Look, I'm sorry for you that you don't get your pet peeve implemented in this game, but Paradox seems to have a very clear view on this matter. And podcat's quotes seem to contradict most of your stated opinions that lead you to believe adding later start dates would be beneficial for Paradox. Of course, you can always pull assumptions out of the blue to convince yourself that 194x start dates would be immensely popular. Nobody can prove you wrong - just like nobody can prove that the flying spaghetti monster doesn't exist - but all the available evidence points in the opposite direction.
"yeah, start dates take a lot of effort (we never managed to balance all the startdates of hoi3 for example) and very very few players ever play anything but the earliest."
HOI3's development process was vastly different than HOI4's, so we're not talking about analogous evidence here. Poorly maintained start dates will, unsurprisingly, be seldom played. We simply don't know what well-maintained post-1939 start dates will get attention-wise.
"Yeah, it doesnt make sense to dedicate resources to more start points. In hindsight 1939 was a mistake to leave in also because only 4% play it. Personally I thought the later startpoints could be a lot of fun in HOI3, but they are not popular enough to focus devs on."
Here he is directly connecting low 1939 usage to his conclusion that post-1939 starts won't be popular enough to focus devs on. Leaving off for now that he is speaking in present tense, and thus doesn't rule out future additions, the
only data points he provides are HOI3 starts (which again were a self-fulfilling prophecy in terms of unpopularity), and current 1939 start usage. As I pointed out in my initial post (none of which you have responded to), that's a bad data point. It's still evidence, but it doesn't actually support the conclusion.
Just as you accuse me of pulling assumptions out of the blue, you're doing the same when the devs haven't actually mentioned any further evidence beyond those two points. If there is such evidence that's great, and hopefully it's actually relevant to the conclusion being drawn, but since these two flawed data points are all we have I just don't find it convincing.
Finally, I'd appreciate if you don't commit a straw man. I'm not saying that they will be "immensely popular," I'm just saying 1939 usage is a bad data point and later in the development process, alternate start dates merit some serious consideration. They don't need to be immensely popular to be economically successful. Grand strategy games as a genre aren't all that popular yet Paradox seems to think it's worth it to make them.