* linux-samsung-p4wifi: Use libressl-dev
* linux-samsung-p4wifi: Update to kernel based on mainline tegra_defconfig

Previously it was based on the config from a mainline-ish kernel for
Android. The remaining Android specific patches have been dropped.
It is now mainline with out-of-tree drivers and corresponding hardware
specific patches/fixes/hacks.

* add some plugins for postmarketos-ui-xfce4
* make start_xfce4.sh start xfce4 properly and add related dependencies

armhf tends to be outdated, just like it is right now. The Alpine
aports have binutils-2.30-r2, and the x86_64 version has that version
as well. But the armhf version is still at -r1, which caused the test
case to fail.

* Add wrapper for i3status for better battery status reporting on N900/bq27200
* New hotkeys "asdfg" to switch workspaces

This is used by shell to show network status in top bar.

libqofono provides the Qt/QML bindings for the ofono, and are used by
Plasma Mobile and nemo/sailfishos.

This new mesa package isn't built for armhf yet in Alpine, but the
previous one breaks plasma mobile right now.

Alpine provides binaries for these now.

This makes sure that symlinks which are needed by rmtfs is created
before rmtfs is started (modem partitions)

- Depend on ofono from device-lg-hammerhead
- From postinstall script of device-lg-hammerhead enable ofono by
  default

On my device the rpmsg0 have DATA5_CNTL under name attribute instead of
rpmsg_name

```
looking at device '/devices/platform/smd/smd:modem/smd:modem.rpmsg_chrdev.0.0/rpmsg/rpmsg_ctrl1/rpmsg0':
  KERNEL=="rpmsg0"
  SUBSYSTEM=="rpmsg"
  DRIVER==""
  ATTR{dst}=="0"
  ATTR{src}=="0"
  ATTR{name}=="DATA5_CNTL"
```

So check for name instead of rpmsg_name.

We need udev running so that we can ensure that the partitions are
correctly symlinked before they are used.

test scripts are useful to enable/disable modem and also do ofono
related tasks, They are installed on /usr/lib/ofono/test directory and
provide much better user experience then the raw dbus-send command.

- Instead of just libqrtr, build all targets of qrtr
- install the qrtr.initd which starts qrtr services before rmtfs is
started
- in post-install enable qrtr service at boot

We pass it to logger anyway where it can be filtered and rmtfs doesn't
produce that much output anyway.

- depend on the firmware-lg-hammerhead-modem and
  firmware-lg-hammerhead-adsp
- depend on the modem-qcom-msm-mainline-common

Currently adsp firmware is not used actually given upstream kernel
doesn't have support for adsp/sound subsystem. But it will be used
eventually.

This includes the firmware files for the Nexus 5 modem and adsp
subsystem. It is currently hosted on my personal server.

This was done for testing locally and is not needed for any
functionality, it managed to slip into the initial qcom kernel package,
but it is not needed.

Upstream suggests to keep remoteproc drivers as modules given they
require the firmware files to be present when they are probed, instead
of putting required firmware files in initramfs due to size
restrictions. We make these drivers modules and load them from
userspace.

Touch screen does not work yet, see the device page for details:
https://wiki.postmarketos.org/wiki/Samsung_Galaxy_Tab_S2_9.7_Wifi_(samsung-gts210vewifi)

Changes:
* device-samsung-klte with both downstream and mainline kernel
  subpackages
* linux-postmarketos-qcom updated to last commit

What works in mainline kernel:
* internal SD card
* volume and home key buttons
* usb network

Due to changes in abuild, our `gcc-armhf` etc. packages did not build
when using strict mode (i.e. `pmbootstrap build --strict gcc-armhf`)
anymore.

Changes:
* Set `CBUILDDIR=/`, so apk can read a valid package index from there
* Directly set `_cross_configure`, so it does not use CBUILDDIR anymore
* Set `BOOTSTRAP="nobuildbase"` to prevent apk from installing
  `build-base-armhf` etc. (these don't exist in pmOS)
* Remove legacy code for lazy reproducible builds that wrapped
  `package()`

matchbox-panel has a keyboard applet, but it was unresponsive.
This was because it sends an event to a running keyboard in daemon
mode. The solution is to start matchbox-keyboard with --daemon in
matchbox-session.

Matchbox-keyboard was not being built as a GTK+2 input method as
thought. This commit correctly enables this functionality and
matchbox-keyboard can now be used as a GTK+2 input method correctly.

* Added gtk+2.0 to depends and gtk+2.0-dev to makedepends.
* Changed --enable-gtk2-im to --enable-gtk-im in ./configure

Add modem-qcom-msm-downstream-common package that pulls in the
dependencies.

qcom_rmtfs now has udev rules to find the storage partitions. My hope
is they can be written flexibly enough to cover all devices and kernel
flavors.