Make parallel installable with Tracker 2.x

See merge request !175

These variables are defined a little bit up in both paths.

Makes those find all executables again.

We do find warnings as we load TrackerDomainOntology first, which
pulls GFile, which pulls GTask, which pulls threads, which makes
glib cry. Sigh.

And make both RSS/files miners keep their data separated (intended
as per 3.0 design, but also multiple writers on a same DB is a recipe
for disaster).

As that data is separated now, and --hard vs --soft makes no difference
in "tracker reset" CLI command, make those be --rss/--files to delete
specific miner data.

This wasn't the case for misc files.

We don't need to look for stores anymore, so clean up that code.

If there's no store, we can no longer kill it, nor track its status/
progress. Drop this code as it's unneeded.

An external change here is that -k/-t no longer allow an optional
all/miners/store argument, it's miners all the way.

Some of those are unused since long ago.

This is not defined properly for the tracker-miners repo. Instead make it
generic, and rely on --disable-fts being passed to the command line tool.
Also drop the stop words check as is a bit of a layering break that is no
longer possible with miner-related CLI commands living in tracker-miners.

This no longer applies since we use meson.

When tracker-extract kicks in the store directory is created long
ago. Just remove this.

This is unused.

There's no reason why this should be installed in plain $libdir.

And update the manpages so they are installed versioned as well.

Matching the executable names

This is now installed as tracker-writeback-3, and referenced as such
everywhere.

This is now installed as tracker-extract-3, and referenced as such
everywhere.

This is now installed as tracker-miner-rss-3, and referenced as such
everywhere.

This is now installed as tracker-miner-fs-3, and referenced as such
everywhere.

Use locale independent printing of doubles

See merge request !176

This will be useful to other users, install it in the default
dbus-1/interfaces directory.

We now use Tracker3 everywhere.

There's no tracker-store process anymore, the store is managed in
process and always available.

We still make those look into the same dconf paths than older
versions, as we want configuration to be in sync.

We are already building the gsettings enums in src/data, no need
to do it here too.

This is no longer used externally. Also remove useless definition
of header files.

This required updating the TrackerDomainOntologies helper in
libtracker-miners-common.

This no longer needs to be installed in tracker's sharedir.

Just make it install into $libdir/tracker-miners-3 as everything else.

We don't allow installing multiple minor versions of tracker-miners
don't make it look like we would.

And make the modules dir agree on the directory name, i.e. no ".0"
suffix, as parallel installability between minor versions is not
observed.

This will be useful later on, as we'll change install locations to
be versioned (with major number granularity, no need for ".0")

sched_setattr fixes for master

See merge request !177

glib#2039 has taught us two
things:
- Even if sched_setattr failures aren't handled as g_error() in
  glib, there will be some kind of warning. It's not desirable to
  extractor modules to indirectly trigger it.
- Since priorities cannot be risen back without special capabilities
  (results in EPERM), it's not that bad to simply allow this syscall.

So simply allow the sched_setattr syscall in our seccomp filter.

This was happening late enough during main() that there were already
non-exclusive threadpools/threads created with regular scheduler
settings. Those settings would be cached in recent glib, creating
disparities that it will g_error() out on later. Those created threads
might however be reused later on by different code (eg. metadata
extraction, directly or indirectly), with the regular scheduling
priorities set.

Given that even accessing GSettings will result in threads being
spawned underneath, there's no better choice than doing this always.
This means the 'sched-idle' setting is ineffective. But this default
should avoid fingers from pointing at us.

This was happening late enough during main() that there were already
non-exclusive threadpools/threads created with regular scheduler
settings. Those settings would be cached in recent glib, creating
disparities that it will g_error() out on later. Those created threads
might however be reused later on by different code (eg. metadata
extraction, directly or indirectly), with the regular scheduling
priorities set.

Given that even accessing GSettings will result in threads being
spawned underneath, there's no better choice than doing this always.
This means the 'sched-idle' setting is ineffective. But this default
should avoid fingers from pointing at us.