Your desired matrix of services is a very good definition.
None exist, as such, nor will, in the near term.
Endless combos of (very) cheap hw and sw can serve your needs in practice - but none will be well supported, or decently documented, or have reasonable chances of being around 5 years from now.
The mid-priced industrial stuff is exactly the same.
The PLCs will be obsolete, the companies will likely be absorbed or transformed, etc.. in 5 years.
Low-none detailed docs, no real expectation of longevity.
Endless cheap linux plcs,
cheap hacks with beaglebone/arduino/raspberry pi/etc..,
Or chinese GSK.
Or chinese clone controllers.
Low-none detailed docs, no real expectation of longevity imho.
Endless cheap motion-controller boards for mach3.
Low-none detailed docs, no real expectation of longevity.
For me, I would use mach3/4 or linuxcnc.
Both are likely to exist for a long time, and go on, with great non-traditional forum support.
If I had to use a packaged hw platform..
a chinese gsk or ccxxx platform would be cost effective with zero support in the real world.
A packaged mach3 system e.g. machmotion etc. would work, and have support, but be very vendor-locked in to a specific company.
Or delta tau, for example. Etc..
Similar applies to linux-based boxed systems with a particular very small company of very limited scope, history, and unknown longevity.
For "supported" systems the CNC refit vendors, or a basic siemens system, would likely be what you want.
All caveats mentioned apply.
There is, imho, very little probability that almost-any of the boxed systems vendors survive above 5 years.
My pov:
The controllers will become commodities - with a very few 1-2 companies in each sector succeeding, mostly due to customer support, docs, pricing either very-low or quite-high/ unit aka lots-of-hobby or very-good-industrial.
And some fairly basic but very functional chinese boxes of very low cost, used in huge numbers.
CCxx offline-controllers around 3-400$, weak docs and weak support, but cheap, and functional for 99% of users.
These types of controllers establish a base price for a basic cnc machine, with gcode and most desired functionality in common gcode, with some limited support for limits/homing/toolchanger/probing/slaving etc.
Mesa/linuxcnc compete with the chinese ccxx and gsk controllers.
So does polabs/machx.
Then, higher-end, e.g. cslabs/machx and siemens stuff, with some high end stuff like maybe machmotion and similar.
Some mesa/linux stuff also competes here.
Multiple hw controllers also exist .. to many axis .. I have no experience but expect they might/could do well short-term in specialist apps- but think their future is limited due to a thin market.
Ie I am sure the multi-axis hw controllers work well, with caveats like everyone - but - trusting any real industrial stuff to a single-vendor solution of low probability of support long-term is crazy.
Thus, a real industrial customer of many axis needs to buy siemens/similar at 40k$-90k$..
or develop plans and know-how and mitigation with a 3000$ multi-axis controller or complex bodge.
It is very likely within a few years one or more box-vendors develops docs, certifications, extensibility, modularity.
If they are then commercially successful, they may become the go-to solution.