Still, the iMac Pro can be considered a good buy if you do like many Apple users and sell it after a few years for a hefty percentage of the purchase price, using the funds to offset its replacement. It's how I progressed from a PowerBook Pismo G3 -> eMac G4 -> Mac mini G4 -> two Intel Mac minis (2006 & 2011) -> Mac Pro tower, only keeping the first gen Intel Mac mini when I moved away from the platform for daily use. Because I recovered at least half the purchase price of each successive unit (again minus the first Intel mini) I was able to upgrade for more or less the same cost as a comparable PC workstation.
Conversely, good luck selling a used custom built PC for pennies on the dollar even if you used top of the line components. Even name brand, quality PC workstations don't have a healthy used market; eBay is full of 2016 era corporate machines that were $800 new, now going for $200 or less.
Well lucky for you: you can add four 4K external displays in addition to the built-in 5K monitor. And you have a PCIe bus coming straight out the back which you can use to plug in GPUs, drives etc.
That being said the machine may not be right for you, but a big corporate shop is probably quite happy to have a bunch of identical machines it can swap out instead of fixing them in situ, so getting the user up and running faster.
Supposedly they learned from the "trash can" debacle and have something more cheese-grater-esque in the pipeline for later this year.
Sounds like this iMac Pro is not for either of us. I'm going the other way: I develop all my code in Emacs on a MacBook (not "Pro") and when I want to crunch, do the crunching on a cluster on my network. I don't even need the 10G ethernet the Mac Pro has, and it wouldn't be fast enough to do the crunching I need anyway. Would I benefit from a faster machine? Sure, but I don't want to carry even a macbook pro any more -- this ultraportable is so tiny I can take it anywhere.
The fact is modularity is the niche, not the other way around.
Most top professionals don't actually upgrade they machines often, because you usually want new everything, and saving a few $1000's every year isn't worth the hassle.
In my mind, there are two main advantages - one is that you can save money by buying only the capacity you need with an option to increase the power later if you need it; and the other is that you can extend the lifespan of the system by upgrading it after a few years instead of buying a new one.
Both of these things don't actually get you a better workstation - they get you cost savings. If you max out the system initially, and when you need something more then replace it with a brand new one, then you get better results, though at a higher cost.
I'd say that the market of $13000 workstations is one where you've already stated that you don't care about savings that much, and you're willing to pay a premium for a better workstation.
Enthusiast home users tend to care about expandability, since they want maximum power at a limited price, and would be willing to spend some time and hassle to save, say, $2000. However, they're not the target market of the premium pre-packaged workstations; they're less likely to pay the "Apple tax" and more likely to build something with comparable power cheaper. On the other hand, companies buying $13000 workstations would rather pay $2000 extra to save time and hassle. If they had a decision "upgrade vs replace" they're going to replace anyway, so the ability to upgrade is literally worthless as they're not going to use it.