Mirror probe does not cull PBR Materials from backside

[ConradAluveaux]

Environment

Alchemy Test 26.2.0.28656348322 (64bit)

Description

Any mirror probe, seem to be able to see the back side of geometry behind the mirror probe, but only if that geometry has a PBR Material on it.

Reproduction steps

1. Make a prim size it <0.3125, 3, 2.5>
2. Make only the mirror side reflective, either by Metallic-Roughness set to 0 or by 255 environment on Blinn-Phong (Tested it with Metallic-Roughness)
3. Make a correctly rotated and sized mirror reflection probe in front of it
4. Observe the mirror working correctly
5. Make the back side of the prim a PBR Material of any kind
6. Observe you are now seeing that surface from behind, on the mirror, instead of your reflection in front of it.

AlchemyTest_sYFCCvWyOw.mp4

[Crexon1]

I wonder if this is something Rye pulled in from the GCP PR Geenz was working on. I noticed something similar with it.
secondlife/viewer#5385 (comment)

[ConradAluveaux]

ooh, i was wondering if it came from some LL build. It didn't reproduce in any other release viewer.
Well at least it is known then. It might be from that issue, it looks pretty similar in behaviors, but odd that my example only reproduces with PBR materials.
Also if the material is double sided, it oddly does not reproduce with it. Very strange...