Fix #300

This is not critical but if we chain multiple migrations we might end up
recreating the backup table while it still exists

Since some thumbnails might be in a broken state, let's regenerate
everything

This now monitors insertion to the cleanup table and batches them.
This allows us to filter what kind of thumbnails we want to cleanup (ie.
only the owned ones) and stop relying on potentially invalid assumptions
when computing the thumbnail path.
refs #350

refs #350

refs #350

Which will store what thumbnail needs to be removed from disk, to remove
the use of a deletion based on a sqlite trigger
refs #350

We're starting to have rather long triggers which don't fit in 2048
characters.

Should fix #356

AutoDeleteAfterUpdate and ShareThumbnail were testing the exact same
thing, so remove one of them

Most of those are not used yet but will be if we ever need a new
version of an index/trigger.
For now they are used as assert only

Which will mark parameters only used in asserts as unused, instead of
spamming the logs with warnings

Some recent changes seem to cause a build failure in release mode, which
isn't tested in any configuration

So we at least check that we can provide a QueryParameter and that it's
correctly interpreted

It's never read from

They were left included during the migration to meson to ensure that all
tests were still run, but now that we made sure they are, there's no
point in including that code in test coverage

We want the tests to explicitely fail if no case matched the given
parameters

When the authority is terminated by a fragment delimiter ('#') or by a
query delimiter ('?') we must not use any potential path delimiter found
later in the URL
Might solve #358

The previous code was fairly broken and wasn't allowing an existing
instance to be replaced. The stop/replace/start sequence was also
broken.
This tries to properly:
- Stop currently running background tasks that use libvlc
- Replace the instance when nothing is using it
- Restart what we stopped

It's worth noting that the fsFactories will be started when required and
don't need a manual restart.
We could do without forcing the parser & discoverer restart, but this
would cause any potentially running scan/discovery to be aborted and not
be restarted until the user explicitly asks for it

Fix #357

Fix #354

This way an application can still store a progress without knowing the
media duration
refs #354

Relying on a media instance means we can't use this code in a migration,
which we do.
While this only moves the problem to File's instances, the File table
is less likely to change than the Media one