Going back to what
@EEE272 wrote - should I be worried that something is wrong with my speakers or amplifier?
I didn't mean to scare you ! Sorry if I did.
I do not think that something is broken, but I still think something is strange about the measurement.
Regarding the mic, you should use the 90degree and point the mic at the listening position to the ceiling. Otherwise reflections are underestimated. Maybe that is what throws me off.
To check your direct sound, you can try a a near-field measure. Here, you can indeed use the 0 degree calibration file and point the mic directly to the speaker. If you get very close, you can also lower the volume. Hereby, you can eliminate some of the room influence to see only how the direct sound looks. That line should be flat. It is also shown in Amir's To be more precise, you can can choose a time window in REW that will then eliminate sound that arrives late and you can focus only on the first arrival. This should also be flat.
@abdo123 regarding your questions. Sorry if I was not clear. The more your room is treated with absorbers, the flatter the line becomes because the reflected sound is eliminated and therefore you have more direct sound. The distance to the speaker has an effect as well due to the shift of direct vs. indirect. Further, above 6K, air attenuation plays a role. This is also why often the slope up to 10K is looked at, as there is some significant drop off afterwards. In consequence, if you draw the line further out, you get the overestimation of the falloff due to the air effect. The 0.4db/Octave are from Floyd Toole's book (he draws it not up to 20K but rather 10K).
But again, I don't think anything is broken! The measurement still puzzles me a bit though but if you could try the above things, I would be very curious to see the results!
Finally, all that sounds well is fine...