Using the field Uploaders within a derivative is misleading and will
produce the Lintian error 'invalid-field-for-derivative'.

https://lintian.debian.org/tags/invalid-field-for-derivative

Default in Bullseye

Update to upstream version '40.0'
with Debian dir 47158da1f32d737e1c7ae7a0733f613511a8e296

And use it to start following the code style in the rest of seahorse.

We're a bit too far off from a consistent style to be able to use a
formatter like uncrustify, but we can start with simple stuff like
trailing whitespaces.

... which is a remnant of the now removed SeahorseWidget object.

Let's start very basic, by just testing if the ..._valid_uri() function
works.

With the advent of HKPS support, some of the code in `SeahorseHkpSource`
got changed quite a lot. With no proper unit tests, it's hard to
guarantee we didn't break anything. With help of the previous commit, we
can now finally start writing tests per backend, so let's do that.

The test actually uncovered some issues with the current HKP
implementetion as well, notably that the first key is ignored. Fix those
while we're at it.

With the removal of SeahorseWidget, there is finally no more awkward
circular dependency going on between libseahorse and common. This means
we can finally start linking backends on their own, which is useful for
tests.

Although it almost physically hurts to mangle our Git history, it makes
sense to start getting rid of trailing whitespace, so we can easily
start enforcing this through our CI, rather than having time-consuming
nitpicks during PR reviews.

Since we're touching these lines already, we can use it to also start
enforcing more and more of a consistent code style.

We've been able to replace it with more modern APIs like `GtkBuilder`,
or by using GTK templates.

This makes sure the dialog shows up on top of the parent window