I have indeed read the thread you mentioned (a while ago).
The theory is basically what was put forward by more knowledgeable people than myself (and I haven't seen a good reason to toss it out,) so maybe instead of here, it could be more productive to take it up with them on the Kerberos boards? It's worth keeping in mind that the sots games are more than the sum of their numbers and even simple behavior can come from a highly complex source. It's also worth noting that there is a good bit built into the games which responds to play style (and it can be VERY hard to change that when seeking an answer through play testing).
Aside (re mockery): I, myself, see snark in response to a dead horse further beaten... But then I think that's something to take up in further detail elsewhere.
I am providing my observations below, if you want to debate them in any detail you know which board to do so on as you've already got an active thread there on that topic. I won't have anything further to add on this topic in this thread on this board (as IMO this is the wrong place).
I've seen that Hiver vs Liir, specifically, have produced some very interesting outcomes, particularly once Liir attained AM and then top end AM engines while my (Hiver) engines were more limited (Fusion).
A specific example that seems to fairly consistently produce the undesirable tactical AI behavior you've referred to is Hiver (me) vs Sol Force (AI). However I haven't done that particular combination in any of the more recent updates/beta (so that data is quite dated).
Outside of those particular examples I hadn't been actively keeping track of which factions were more lemming train as you put it and which weren't, I just know that I didn't always experience that particular problem as some people (ie not just you) apparently did/do. (ie the issue isn't any less real imo, just less global/all encompassing and based on the thread referenced no easy solutions available.)
More generally:
I *think* at least one of the more recent updates did some tweaking on the tactical AI as I don't remember seeing fleet components splitting into more than 1 attack and 1 fleet support.
I have seen the AI split it's attack component into two and go after 2 different targets.
Split or not, their targets have sometimes been planets (ie the lemming train), sometimes assets like stations (within a sector usually width or 2 in depth), sometimes platforms (quite nearby usually), and yes, sometimes even my ships. (The order there is not meant to confer any particular weight, best consider them fairly equal.)
In some cases it was less than obvious what their target was, in others it was pretty clear it was my ships or something other than a planet.
In some cases they wanted to get to a certain engagement range which was outside of where I wanted to be (the ships would turn back once my ships had moved far enough away).
I've seen some unusual behavior if I had little in the way of stand off weapons (which isn't common) as well... I've seen them run at my ships (over thrust to get in range) as well as overthrust to break away (both no obvious target and obvious target).
I've even (rarely) seen an attack component change targets... for example if one component managed to put me on the ropes so to speak (damaged ships, lost sections). I usually lost ships shortly after.
(That was all observations while defending against an AI attack. Unless otherwise noted these weren't one off or exceedingly rare events.)
As far as 3rd party observations (ie global chat)... I've seen people talking about their empires being utterly shredded by the AI. Tidbits about how the AI was wiping out their fleet before moving onto the planet/stations. Granted that was mostly from new players of the game or people that took an extended break (as little as 4 or so months) and coming back just recently. I've seen some older hands (besides myself) seeming to have some rough games more recently. I know mine have gotten more interesting though I still know what to poke on the AI to kill it right quick.
Overall I stand by my original shorter observation:
Right now I find the AI to generally be fun if a bit over predictable. AI seems to be getting dangerous if you goof up or have a sufficiently bad start.