Fixes #7363

closes #7452

Breaking a public key of a person can be used to "block" receiving posts
from this person on the pod. So we should handle that case better and
not just trigger many retries for something that will fail again.

closes #7448

This is now broken since more than half a year, so I think we can just
remove it.

Podmins shouldn't use this anyway, because it does bad things, like
replacing NULL with an empty string when editing something. If we really
need an admin interface for something, we should add it to our own admin
panel at /admins/dashboard.

Fixes #7027 and #6413

closes #7440

Fixes #7249

closes #7420

Fixes #7438. The regression has possibly been introduced by #7207.

closes #7441

The previous version logged the encrypted private messages of the new
protocol, which only makes the logfiles bigger without any benefit,
because it's encrypted anyway ...

closes #7435

closes #7419

Here we also set autoDispose to false. This is an option that was introduced
sometime after our last bump. Default is true and it disposes timeago object
if it is not in DOM. But that is the way we initialize timeago objects, therefore
if we have autoDispose=true then they will be disposed before they inserted in DOM.
So by using autoDispose=false we enforce previous behavior that worked fine for us.

closes #7429

fixes #7427

[ci skip]

closes #7414

Fixes #4991

closes #7367

"should see JSON"

closes #7410

closes #7404

closes #7403

closes #7401

Since 4.2.3 sidekiq allows to disable its own session cookie for the
case where the app provides a session already.

closes #7395

These lines might have been necessary some time ago but now they are making things worse.

closes #7374

closes #7376

Fixes #7321.

Fixes #6248, supersedes #6244.

closes #7383

closes #7385

Pass destroyModel from post control view to post view
and don't listen to the remove event in the post view
anymore. The `remove` function will be called from the
base view.

Fixes #5445