Low video resolution and aggressive downscaling on ESS 26.1.1 (Element Call 0.16.3) before livekit

### Steps to reproduce

​1. Where are you starting? What can you see?
​I am starting a 1:1 video call between Element X (on a smartphone) and Element Web (on a PC). Both devices are on a stable, high-speed 330 Mbps fiber connection. I can see the call interface, but the initial video quality is very blurry.
​2. What do you click?
​I initiate a video call from the smartphone. Once established, I click on Settings > Help & About > Show RTC Stats in Element Web to monitor the incoming stream's technical data.
​3. More steps…
​Wait for the call to stabilize (approx. 15-20 seconds).
​Observe the resolution ramp-up in the JSON stats.
​Perform a "Camera Flip" (switch from front to back camera) on the mobile device.
​Observe the temporary resolution jump and the subsequent downscaling.

### Outcome

What did you expect?
I expected the video quality to reach 1280x720 (720p) @ 30fps within a few seconds and stay there consistently, given that the server has plenty of headroom (Xeon E5-2690 v3, 4 CPU cores / 2GB RAM limits for SFU) and the network bandwidth is more than sufficient.
What happened instead?
The call starts at a very low 180p. It takes about 15 seconds just to reach 360p/540p. The resolution seems to hit a "ceiling" at 540p and refuses to stay at 720p. Interestingly, flipping the camera causes a brief jump to 720p, but the system aggressively downscales it back to 540p or lower within seconds, despite 0% packet loss and low jitter.

### Operating system

Windows, Android

### Browser information

Edge (latest ver.)

### URL for webapp

_No response_

### Will you send logs?

No