As far as I understand, Mozilla's main point is how these technologies
prevent Firefox's core from easily evolving, especially performance-wise.
An exception was made for Flash but it is easier to maintain NPAPI working
for Flash than for any plugin. That said, I had rather see Flash support
dropped today (the same holds for EME, of course). Flash support will be
dropped soon since Adobe "will stop updating and distributing the Flash
Player at the end of 2020":
https://blogs.adobe.com/conversations/2017/07/adobe-flash-update.html
As for "add-on developers who will switch to other browsers", they would
switch to Chrome/Chromium, which unfortunately dominates the market:
http://gs.statcounter.com
In fact, I believe most add-ons for Firefox exist for Chrome. Their
developers love the API change. Since it is that of Chrome, they have far
less porting work. For those only developing for Firefox or Chrome, they
will probably start supporting the other browser (again, it is little work)
and I see little reason to stop supporting the browser they come from.
Old unmaintained add-ons will indeed break. But it looks like a reasonable
price to pay for improved performance. After all, if nobody wants to
maintain the add-on, it is probably not that useful.
The reasons you report to drop sound support on GNU/Linux without PulseAudio
("requires more work to support than Pulseaudio", "focus on other features in
the browser"), combined with the overwhelming dominance of PulseAudio in
modern GNU/Linux distributions, make a good argument, don't they?