I disagree on the "tough" part. In my opinion this is less difficult than DOSbox/Wine on Linux or DOS subsystem in Windows 32 (and Itanium editions of Windows 64). It is more of the problem how much energy to spend on scoping the required area of backward compatibility and preparing/verifying test cases.

The initial step is already done in form of libconsensus. It is a matter of slightly broadening the libconsensus' interface to allow for full processing of compatibility-mode transactions off the wire and old-style blocks out of the disk archive.

Then it is just a matter of keeping track of the versions of libconsensus.

To my nose this whole "segregated witness as a soft fork" has a strong whiff of the "This program cannot be run in DOS mode" from Redmond, WA. Initially there were paeans written about how great it is that one could start Aldus Pagemaker both by typing PAGEMKR on the C> prompt (to start Windows) and by clicking PageMaker icon in the Program Manager (if you already had Windows started). Only years later the designers admitted this to be one of the worst choices in the history of backward compatibility.