Interesting idea, but it would cripple Apple's efforts to transition away from x86 architecture, as they'd need to provide x86 support for longer. My bet would be that the whole native x86 macOS thing has no more than 5 years left.
x86 macOS support should be around for at least 5-7 years (likely longer), but that does not mean a macOS version released 6+ years from now will have x86 support. It might only be supplemental update for an older OS.
A lot depends on the next 2 year roadmap and/or if they can make a MP7,1 style rival chip for high end needs, or if they abandon the "PRO" market once again.
Bigger issue is the amount of third party who will depart macOS in the next 2+ years. The iOS/iPadOS devs and companies with cross-platform apps will stay. Hard to see some of the smaller ones starting from scratch if Rosetta 2 isn't still offered natively in latest OS 5+ years from now.
Why would it be slow, unless you’re transferring huge amounts of data back and forth with the host MP? CPU, RAM, even storage etc. is all self-contained on the MPX. It would be just like a Phi running its own OS, and you telnet (or whatever) to it via PCIe. No complaints of Phi or other coprocessors being slow.
Seriously though, I think the era of dual boot is over. It would require insane engineering to accommodate a dual platform ARM/Intel hardware thing. How do you share buses, memory, timing, T2 chip, etc etc without compromising one or the other or have constant crashing? The cost would be enormous.
It would probably never be truly compatible with all PC software, either.
A decent PC can be had dirt cheap if all you need is occasional Windows access.
If you really need a powerful Windows workstation, or a gaming rig, then you're probably building your own, or buying something like an HP Z series.
Increasingly, you may not even really be working on a desktop but in a hosted VM in the cloud somewhere, or accessing your work PC via Microsoft Remote Desktop. Or your tools are all web-based or cross-platform anyway.
I think all the trends point to a greatly decreased demand for dual-boot. Is it convenient and cool? Sure, but think about the trade-offs from Apple's point of view.
Dual booters (talking pros, excluding enthusiasts) are a shrinking slice of the market, and the cost to support them probably does not make sense anymore.
By retaining Intel / x86 hardware, it hobbles their ability to make a clean transition to a new architecture. For laptops, it means stlll dealing with hot, power-hungry Core chips, vs being able to get the big compute-power-per-watt / battery life boost that the A-series platform will give them.
It's not like 2006, when Apple was trying to cement its decade of recovery from where they were in 1996. Maybe they needed to tout Windows compatibility as a way to lure corporate buyers back, but now that macOS / iOS is 20 years old, IT departments are increasingly flexible about giving you a choice of machine. In many cases iOS is even preferable because of its locked-down nature (when developing and distributing corporate apps using managed devices).