Correct way to render Camera in VR-PostProcess

I want to have a post effect (PPv2) which uses the default rendered camera image and a second one which is rendered with a different CullingMask.
In the past I did a manual Camera.Render() but this is a recursive call and forbidden in the SRP.
What is the most straight forward way to do an equivalent.
Had it working on the Default pipeline as well as the HDRP. Just the URP is complicated/unclear.

I tried to use a second camera which is rendering to a render texture.
This resulted in doubled pictures on each eye, so the coordinates are not set up properly.
What is the correct way to sample the manually rendered texture ?
(And is there anything special in setting up the Render texture ?)

Thanks in advance for any help or insight

 

But how to get the ScriptableRenderContext needed as parameter ?
(Inside the Post Process)

But then would the image not be stereo
Basically I need to render the VR Cam two times with different Settings and use both rendered pictures in a PostProcess.