I am not entirely certain of this but it looks like a wonky race between
the vt service prerequisties starting and us going to the vt resulting
in the getty not actually starting properly.

we only care about the icon on the desktop

aspect of wallpaper changed it seems

this turned out to be a race between udevd acting on actions taken
by calamares' mount cleanup code and sfdisk run by kpmcore.

in lieu of an upstream fix we'll force a settle before all sfdisk calls.
these should be super fast when there's nothing to settle.

this change should make the partitioning test pass reliably again

This reverts commit 3552fa99.

it looks like my logout fix actually solved the problem I was seeing there

in my testing this seems to happen automatically but best be precise
here

something is wrong there and the vt6 is broken afterwards.
I am not sure the logs will help but they just might.

I also don't seem able to reproduce this :|

- previous needle is now 5.15
- drop 5.14, no longer in production

sddm and sddm idle split into 5.15

also adding sddm-2, this is a divergent 5.15 wallpaper but with 5.16 blur
changes, so it's different but strongly related to 5.15. keeping this
to not regress current installations

also drop idle vs idle-unstable. I haven't checked the history but it makes
no sense. we have TYPE tags, they should be used to select a given
needle for a specific type

also unlimit sddm from unstable. no use only having it tested there TBH
only gives a headache once user/testing roll over to new wallpaper and
we assert this kinda elsewhere anyway

This reverts commit 17f41773.

fix should be merged now

it trips up stdout parsing I think. it parses stdout :|

it's kaput
https://progress.opensuse.org/issues/52235

labels changed

lest calamares fails to parse output :O

hijack sfdisk calls and route their output through tee into a log we can
archive

This reverts commit e9207a23.

the previous commit to attempt sleep only after welcome didn't achieve
much. still working under the assumption that this is a race, let's
sleep all over the place

This reverts commit 04676fd9.

this again failed unreasonably. I am not sure what introduces the delta
but it's fairly subsantial

this is a try to prevent the partitioning failing on obscure fdisk
errors without useful error output. I believe this is because of a race
when initializing the modules, but find it hard to proof that theory.
if a sleep really fixes the problem in many runs then we can be fairly
certain that it is a race.

why they are called all sorts of different names is anyone's guess. this
like all the other updates simply adjusts the areas of the final cala
page so they exclude variable portions of the name strings

was is misalinged because I updated the needle and the panel
now has a show desktop appelt as well

as of late mouse_hide of os-autoinst always triggers the mousearea of the
edit box, so sometimes its enabled, sometimes its not. very annoying
but easiest to just not assert the right most element

unfortunately we can't control where the mouse gets hidden to, so, meh.

this should make the needle slightly more reliable

now tests
- first login cold (i.e. after reboot)
- second login cold
- third login warm

for the purposes of the first login a  bunch of standard dirs are ripped
out of the user, since the user was alrady adjusted for testing scenarios
so it has a fixed wallpaper etc.

there's now also a sleep after finding sddm to ensure the system has
quieted down a bit and doesn't use resources which would go to the login
otherwise

requested by plasma dev. might be worth extending this a bit to cover
time on clean profile vs. existing profile

to increase the matching change, due to how the slideshow starts
currently the slide appearance is a bit wonky

colors are a bit different, but most important the aspect has changed
because we no longer have a summary that excessively resizes the window

now that we no longer use a summary it's not a given that the button is
highlighted

just in case...

summary has been kicked to the bucket in calamares-settings