1, could you show the media-info of your rendered HDR-file

2, why do you use rec.2020 ste 2084 2000 nits for your view transform and master display

3, for what reason do you choose that deintelace method? do you really understand it?

Youtube takes a long time to render 4K videos and I assume HDR is even longer. They'd probably put up the SDR version first and then the HDR comes later. I'd try again tomorrow as Musicvid said. Make sure the device you're using supports HDR like a smart TV or windows 10 youtube app on an HDR display.

1, could you show the media-info of your rendered HDR-file

2, why do you use rec.2020 ste 2084 2000 nits for your view transform and master display

3, for what reason do you choose that deintelace method? do you really understand it?

Hi Yellandkeil. For your second question: These are the settings I found I had to use according to the Elgato site. I don't know anything about them, I am just blindly following guides in the web. For your third question: No, I don't understand a single thing in the settings, I just follow guides. I'd be happy to know more if you want to explain.

HDR10 is mastered to 1000 nits. If your metadata is conflicting with Youtube's standards for HDR video it won't work. I'd try rec 2020 ste 2084 1000 nits and also see here for more info:

https://support.google.com/youtube/answer/7126552?hl=en

VEGAS' default HDR settings work fine with youtube. If you're rendering to one of VEGAS' HDR presets (make sure you aren't using a custom preset), youtube will recognize it as HDR once given enough time to process the file.

@MarshallEden

To point 2 you've got the answer I think.
To point 3, that's a long long story. But we make it short: 

Traditional video signal has - e.g. PAL-standard - 720x576_50i, namely there's 50 motion pictures to be transmitted per second through the air or cable. 

An impossible mission that time!
Because: 720x576 was a tremendous resolution which no equipment had the bandwidth to cover. 
Apropos: 720x576 = 720 impulse clocks horizontal and 576 scan lines vertical. 

Smart engineers found a "trick", they split the scanning in 2 sequences: 720x288 odd lines as "Upper Field" and 720x288 even lines as "Lower Field". 
This special scanning/signal comes through the air/cable with same frequency but half bandwidth into the special receiver, say TV. 
And TV takes the same trick inversely so that our eyes can't distinguish whether it's 50 half fields or 25 whole pictures. We see fluent motions.

This technique is called Interlaced Scan, very brilliant but suffering lost of picture quality. 
Today, our hardware has every bandwidth more than enough. 
Interlaced scan vanished. Its legacy remained. 

At playing such videos, modern hardware doesn't "combine" the 50 half fields, it treats them as 50 whole pictures, with technical compensation what we needn't study. 
And this compensation is called in Vegaspro:
Smart Adaptive (GPU only) 

Conditions to use: 
1, your GPU can do it
2, your output is 50fps
3, you really have interlaced source material on the timeline
4, you only need it by rendering, not editing. 

Tipps: Use "Interpolate Fields" as temporary deinterlace method in THIS case. 
(In other cases it's over again another long long story.)

To point 1, you need give me nothing any more as you've got point 2, and for your info here's an utube streaming dataflow: