Pro users seem to divide into two sorts: those who want the horsepower and basic connectivity (ethernet, thunderbolt etc), but don't really make use of serious on-board expandability (depending more on network storage and that sort of thing); and those who want to make use of the on-board flexibility for storage, specialist expansion cards and the like.
With the trashcan, Apple took the path of assuming its customers were primarily of the first sort, but with the cheesegrater they backtracked and embraced the second sort.
Apple never really backtracked. MP 2013 --> iMac Pro ( stopgap 2020 iMac 27" ) --> Mac Studio
There was a hiccup stopgap on that path because the Mac Studio came out so late in the transition, but those were all :
i. bounded literal desktop footprint of the Mac Mini's 7x7 inch box.
ii. About 400W power supply.
iii. Larger GPU VRAM option
iv. One, and only one, internal drive
v. $2,999-$4,999 entry point
The MP 2019 gave up on i-iv but did not on v. Apple lopped off the 2006-2012 Mac Pro users that were in that zone ( plus the ones that wish MP was still in the $2-2.5K zone. ). That price increase
STILL thinks that customers are primarily of the first sort. Otherwise they wouldn't be leaving a large block of folks with more limited budgets behind to set the 'foor' price to the $6K ( a 100% increase. 100% increase you
NOT going after exactly the same group. It is a subset. ).
Apple shifted to a different user segmentation at least as much "backtracked". A Vega II / W6800X Duo all by itself cost about double what an entry MP 2010 did. That is whole new 'zipcode' for a price range.
All of which gets to the core problem the thread's start inquires about; the "We are a bigger and/or 'more important' group than those folks" issue. Apple was trying too hard to herd too many folks into one bucket to the point there is lots of ranker about what the boundaries of the bucket are. Much of the 'over the top' anger about the 'Trashcan' is about the self perception that this is the larger/'more important' group. Apple just likely has a different viewpoint. Mac unit sales and revenue from 2012-2019 went up. Did they miss out on most users? Nope.
Apple Silicon has 'doubled down' on some of those first four aspects that Apple has stuck to since 2013. The Mac Studio is just taller in the same footprint. The whole line is focused on keeping power down; not draining the maximum amount possible from a wall socket. Larger VRAM is being spread to the whole line up! The security intertwined with the primary boot drive has just gotten deeper.
Some folks took the MP 2019 as some complete 180 degree turn toward hypermodular infatuation by Apple. It allowed optional drives (not sold at all in BTO ) but the T2 SSD was firmly there. There was better 'off-the-shelf' GPU coverage for AMD cards , but Nvidia was gone and much of the GPU coverage was driven by dGPU in other Intel Mac systems. The 'off-the-shelf' major driving force was the UEFI that Intel chips needed anyway (not particularly driven by Apple). The lack of an iGPU in the CPU was Intel's call; not Apple's. The TDP targets for the GPUs ... AMD's call; not Apple's.
Don't force folks to buy more than one GPU really isn't about hypermodular love.
Don't lean too hard on Thunderbolt is giving to some internal expansion, but not making hyper internal flexibility the total end goal. Thunderbolt was in no way optional on a standard MP 2019 (MPX connectors , etc. ) .
Commitment to Extremely high bandwidth is not necessarily '==' standard PCI-e slots.