With the ARM-Mac cresting the horizon, the line of Macs that immediately precede their arrival are likely to be the pinnacle of the current 'classic' Mac.
Which last happened in 2012 with the launch of the radically-redesigned Retina MacBook Pros (which were accompanied by updated "classic" MBPs). As far as I can remember it was likewise with the PPC-to-Intel shift (in so far as there
was no upgrade path for the Powerbook G4 at the time, which was part of the reason for the shift).
Unfortunately, since then, most of Apple's 'radical' changes to the Mac line have consisted of letting the previous models get laughably out-of-date, with CPUs/GPUs a generation behind those available in PCs. See: 2016 MBP, 2018 MacBook Air, 2018 Mac Mini, 2019 (maybe) Mac Pro, 201? 12" Macbook...
The way things are going, the 2017 iMac could turn out to be the 'pinnacle' that you mention... but it would be a pinnacle without the benefit of 8th/9th gen processors with 50% more cores that could make a big difference to some workflows...
I really don't envy anybody who needs a new iMac right at the moment - I'm alright because I got my 2017 iMac in, well, 2017, and although I'm curious, I wouldn't expect to upgrade for at least another year (and I might be more inclined to consider an ARM-based Mac than others). The 2017 is a solid machine, but better chippery is available in 2019...
That's if the ARM rumour turns out to be any better than the "New iMac last October" ones. I'd put that at 50/50 - personally, I think its more achievable than some people think (modern apps written in high-level language and using OS frameworks for graphics and acceleration should be trivial to convert and forthcoming changes such as the end of 32 bit are going to kill off a lot of 'abandonware' anyway). Still, the #1 way that Apple could stuff it up is to push it too fast and face people with a choice between an ARM machine - which isn't going to run all of the software you need on day 1 - or a 2-3-year-old design (at the same price as when it was brand new).