outer RC slab - dilatations for crack control?

Depending on your environment, you might want to prevent gravel falling into your crack inducer with a sealant or filler board. With temperature fluctuations, the sawcut/ etc. can narrow with local spalling when gravel fills the gap. Crack inducer/ sawcut should be 25-33% of the slab depth.

Have you put dowels across those joints or something else? Intersecting dowels lock up, so you usually need to interrupt the dowels in one direction at intersecting joints and allow dowels in the other direction to continue.

Yes, there will be dowels in the middle of the slab thicknes.

Would you do the sawcut across the part of the slab that is on the different level than the rest of the slab as shown bellow?

any ideas?
not my field...

That arrangement is a bit unusual in slab pavements. Your lengthways sawcut joint needs to continue through the dropped slab.

I wonder whether you will have stress concentrations at the corners of your dropped slab.

Here are a few rules of thumb I follow when laying out control joints:

1. Max panel side ratio = 2:1 max
2. Joint spacing = 3*slab thickness in inches = joint spacing in feet
3. Avoid narrow or sharp panel corners where multiple joints meet. (Keep the angle between adjacent joints 60 degrees or more).
4. Joints should intersect at a common point. Avoid "dead end" or "tee shaped" joint arrangements in plan view
5. Provide joints at any changes in slab thickness or abrupt changes in elevation.
6. Provide joints to the corners of around any slab penetrations or protrusions (openings, columns, pits, equipment pads, etc)

I am sure there are more good tips out there, but those above should get you close.

Side note: I had to Google "dilatations" and based on the definition I found, I cannot understand what you are intending it to mean. Is that some local structural jargon?