udev rules and manpages in section 8 were installed into $out/usr

Results in the following derivation output diff:

 lib/systemd/system
 lib/systemd/system/multipathd.service
 lib/systemd/system/multipathd.socket
+lib/udev
+lib/udev/kpartx_id
+lib/udev/rules.d
+lib/udev/rules.d/11-dm-mpath.rules
+lib/udev/rules.d/11-dm-parts.rules
+lib/udev/rules.d/56-multipath.rules
+lib/udev/rules.d/66-kpartx.rules
+lib/udev/rules.d/68-del-part-nodes.rules
 sbin
 share
 share/man
@@ -79,25 +87,14 @@
 share/man/man3/mpath_persistent_reserve_out.3.gz
 share/man/man5
 share/man/man5/multipath.conf.5.gz
+share/man/man8
+share/man/man8/kpartx.8.gz
+share/man/man8/mpathpersist.8.gz
+share/man/man8/multipath.8.gz
+share/man/man8/multipathd.8.gz
 usr
 usr/include
 usr/include/libdmmp
 usr/include/libdmmp/libdmmp.h
 usr/include/mpath_cmd.h
 usr/include/mpath_persist.h
-usr/lib
-usr/lib/udev
-usr/lib/udev/kpartx_id
-usr/lib/udev/rules.d
-usr/lib/udev/rules.d/11-dm-mpath.rules
-usr/lib/udev/rules.d/11-dm-parts.rules
-usr/lib/udev/rules.d/56-multipath.rules
-usr/lib/udev/rules.d/66-kpartx.rules
-usr/lib/udev/rules.d/68-del-part-nodes.rules
-usr/share
-usr/share/man
-usr/share/man/man8
-usr/share/man/man8/kpartx.8.gz
-usr/share/man/man8/mpathpersist.8.gz
-usr/share/man/man8/multipath.8.gz
-usr/share/man/man8/multipathd.8.gz

also switch to python3

Since this package symlinks pyqt5 in its own tree, propagating pyqt5
is detected as a conflict by our python setup hook.

otherwise we get symbol conflicts, once we link it against applications using
openssl

- fixes build against boost
- also enable DoH

This version is not yet released. However given that python2 will soon
go end-of-life (without security updates), this seems like a good move.
The package was also lacking proper qt wrapping and unusable before.

Makes it possible to use ANT+ sensors using an ANT+ USB transceiver.

In hopes of working around
NixOS/nixpkgs#60845. This fetches a gz archive
instead of xz and also fetches the archive from a different source,
hopefully avoiding the issue (whatever ist is caused by).

In addition to that, I think that building directly from VCS is
generally cleaner and more flexible for the following reasons:

- It cuts out and unnecessary middle step.

- It makes sure the version users install is equal to the version users
may have vetted.

- It makes it easy to develop patches or bisect changes by simply
checking out a different rev.

- It avoids using upstream-provided "binary" artifacts like those
generated by autotools.

Changelog: https://github.com/Rigellute/spotify-tui/releases/tag/v0.11.0

This adds the hicolor theme to XDG_DATA_DIRS for rofi and moves all
wrapping to rofi/wrapper.nix.

The wrapper will now always be generated, even when a theme is not provided.

This only formally fixes the name of the compiler. Theano appears to be
working on darwin after this.

croc: 6.4.7 -> 6.4.8

dbeaver: 6.3.0 -> 6.3.1

spotify-tui: 0.10.0 -> 0.11.0

nco: 4.9.0 -> 4.9.1

palemoon: 28.7.2 -> 28.8.0

mixxx: 2.2.2 -> 2.2.3