Kindle Scribe: Fix KindleUI L and R orientations when restarting KOReader book in landscape

[mergen3107]

Fixes #11743

It turns out that KOReader can only digest U or D orientations.
These are 0 and 2 in the list below:

self.screen.DEVICE_ROTATED_UPRIGHT =  0 
self.screen.DEVICE_ROTATED_CLOCKWISE =  1 
self.screen.DEVICE_ROTATED_UPSIDE_DOWN =  2 
self.screen.DEVICE_ROTATED_COUNTER_CLOCKWISE =  3

1 and 3 cause the touch to be misaligned with the visible rendered UI by 90 degrees if a KOReader book has been exited in KOReader's "landscape" orientation, and I don't really know why. Probably, because KOReader "rotates" the UI (or touch layer?) by 90 degrees when it is in "landscape" mode.

Having U/L and D/R in the same groups worked well.
If Kindle is started in L, KOREader will greet the user in the same orientation, if the KOReader book has been left in that "landscape" orientation.

---

This change is 

[mergen3107]

Tested well in Kindle Scribe.

I don't know if Kindle Oasis 1, 2, 3 are victims of the same trap.

If so, let me know, and I can make another PR.

Thanks!

[mergen3107]

@Frenzie
If @NiLuJe approves, it is ready to merge.

[Frenzie]

I'm not sure I quite understand. This means the system is transparently turning everything around and the problem lies in trying to handle it?

[mergen3107]

I don't know exactly, but it seems system doesn't like when KOReader rotates it to 1 or 3.

There are two weird things I still don't understand:

1. When I open Kindle UI book from U, and make it landscape, the book rotates to R physically, but the lipc still reports U.
   Then if I rotate by gyro back and forth, it rotates and reports L/R respectively.

2. If I then quit to Kindle UI, it physically only stays in U. It reports the old L or R. This is exactly when KOReader was broken if I tried to start it here.

If I rotate to U/D in Kindle UI, it reports U/D respectively.

So I found a very narrow use case when KOReader broke.

This PR fixed it.

[mergen3107]

Regular modes (U/D) didn't break, so all good.

[NiLuJe]

Sounds.... weird, but whatever ^^.

It does make some mild sort of sense that they'd be handling the orientation swap somehow :?

[mergen3107]

Yeah, I am still confused every time I try to implement hardware rotation as suggested in #11269 (see my comments in the end).

It seems like system is trying to override the orientation at least upon exit.

I'll try to iron out these issues later too

[mergen3107]

Should I wait until #11807 and co. are merge and update my PR?

Or do you guys want to merge this first?

[NiLuJe]

That shouldn't conflict, I think?

[mergen3107]

Yes, they don't conflict.
I just needed to manually apply my PR again after successful tests with your PRs.

I don't know whether it needs to be rebased for this PR, for example, after applying your PRs. And vice versa

[NiLuJe]

Yep, I meant "conflict" in those very git merge terms ;p.

The patch hunks look separate enough that it shouldn't be a problem either way ;).

(If in doubt, merge this one first, I'll rebase; the other needs a base bump anyway).

[mergen3107]

Thanks!
@Frenzie could you please merge this PR?