First of all, 'local' is non-POSIX, but it is a really good safeguard
against hard-to-find bugs. However, doing a local + initializing combo
like `local X=$Y` in dash is error prone. If `Y=1 2` will get an error
since dash will expand $Y so we get `local X=1 2`, but it will treat
the "2" as another variable to be made local, which isn't what we
want. Hence, let's declare variables local and initialize them in
separate commands, which is safe.

It has been moved into a file we cannot source.

Earlier, if it was set to e.g. 'removable-usb' and no USB storage was
connected, $whitelistdev would be empty, which is interpreted like
all devices are ok by the rest of the code.

But we add an option --gpt-only to get back the old behaviour.

It seems Tails may have been able to list GPT partitions labelled
"TailsData" on hard drives (!) as valid persistence volumes...

error() uses the new fd if souring fails, which happens before main().

While we're at it, also remove some stderr null-redirections since the
output can be useful in the log files.

The check cannot become true ($dev is unused) and wouldn't be
usefull any way since one might have mounted a partition manually
and then want to use this script to enable its persistence.

Until now, the value passed to this option was, well, ignored.

This branch is about upgrading Tails to live-boot 3.0~b7-1 and live-config
3.0.12-1, and may not be compatible with any version pushed to sid after the
next Tails freeze.

Let's keep the >>stable pinning for unstable though, it's a sensible fallback
for branches of ours that don't ship custom live-* packages.

They are renamed, "sometimes",
and failing early makes it easier to detect and understand
when it happens.

    * /lib/live/boot/cmdline.sh was renamed
      to /lib/live/boot/9990-cmdline-old
    * /lib/live/boot/misc-helpers.sh was renamed
      to /lib/live/boot/9990-misc-helpers.sh
    * Cmdline was renamed to Cmdline_old

Recent live-boot mounts the read-only live image in there..

They have been removed from our feature-live-boot-3.x APT suite too,
but are still in the devel one,
that is used too for the topic branches builds,
so we need APT pinning to fetch them from sid.

The documented way for "external" contributors to add custom packages is to put
them in chroot_local-packages, and once we pull we import any such package into
our APT repo and rewrite the history appropriately.

Also, the ability to add packages in there and not see them in "git status"
makes it very easy to build tainted ISO images with non-standard packages,
which makes some of us fear can lead to hard to debug situations.

Conflicts:
	debian/changelog