> > The correct solution: Cache tab history completely [...]
This is not a solution for the majority of users. The problem with keyboard shortcuts is that most users won't know what happened. All they'll know is that suddenly they're not on the same page any more, and then won't realize they've navigated backwards. Therefore they won't think to navigate forwards again, and they won't get back to their form data, even if it is cached.
This happens to me all the time in Emacs. I'll press something by accident, make major unintended changes to my file, and have no idea what I did. Fortunately, Emacs has a way to look up key presses, but this is a relatively advanced feature that typically doesn't exist in other tools anyway.
Whenever you have a feature with zero discoverability (e.g. essentially all keyboard shortcuts), you have to be very careful to avoid the "what did I just do?" phenomenon, even if you provide ways to get back to a previous state. I personally think this change by the Chrome devs is extremely reasonable in that regard.
Edit: I realize that these changes are aggravating to users who actually use them on a regular basis. However, please also realize that they are not zero-cost, because they do impose a constant and non-trivial burden on everyone else (especially in cases like backspace which are relatively easy to hit on accident).