Because current add-on design do not work with multi-process!!
They are not ending the add-on, they must be migrated to the new API. Sadly many add-on are abandoned and will not be migrated. Others will not be allowed to do some functions, almost all of then must be rewritten. But the current add-on have fatal flaws and are doomed sooner or later.
There are several problems with the current^WOLD add-on layer
1- Old add-on have too much access to the firefox internals (security, memory leaks and performance problems)
2- Old add-ons do not know how to work with multi-process and mozilla had to simulate a pool+lock for then to work (big lock, performance problems)
3- Migrating code from Servo to Geko would break many add-on unless there are many compatibility layers (performance and code maintenance problems)
1-They could swap unsafe parts slowly and break many add-on on each release, forcing a slow and never ending add-on update cycle. It is much easier to just swap the API and warn that everyone must rewrite.
2-Add-ons need to be fixed or else the browser will not really use the multi-process well and worse, may be even slower because of the big lock. If they remove the compatilbity layer, add-on stop working, but postponing the removal will keep the browser slower too and the add-on may never be updated (multi-process is already a several years project and all add-ons where flagged to be updated, but many are just abandoned). So wait more is not a solution, they need to be rewrite
3-No one wants layers over layers, it is a maintenance hell, specially because the add-on have access to almost everything. Migrating to a simpler API make mozilla job much easier, firefox safer. Add-on will have to be rebuild and developers need to learn a new API. They will also be unable to do some things they can right now, but on the good side, it will much easier to port add-on between chrome and firefox and the add-on can be run in separated process, so bad add-on will be easier to stop and control.
Yes, i too would like to keep all the add-ons, but between a fast browser with fewer add-ons and a slow one with many outdated add-ons, i prefer the first one. You can not complain about firefox being slow and also complain about keeping old add-ons. To fix one, you need to fix the other too! and you can not delay this, market share is shrinking due firefox being slower.
They were making changes slowly, as they were mostly doing the last few years, and try to not break the add-ons, but you can not postpone a big internal change forever and now it is time to drop some old features, like NPAPI plugins and the old add-on interface
What i hope is that mozilla is now more open to some features, as some features will be blocked to add-ons, mozilla need to be more flexible on certain features. The google design model ("only allow features that at least 80% of people use") is bad for firefox, as many of the users of firefox are the 20% of excluded people in chrome