[DaVinci Resolve] Improve IBKeyer integration

So  ⁨@CorridorCrew⁩  just released the 'Corridor Keying' system, I got to to be honest from a workflow standpoint I can get a better key faster with an Image Based Keying system from Nuke. Corridor Key Video: [   • It Took Me 30 Years to Solve this VFX Problem  ](https://www.youtube.com/watch?v=3Ploi723hg4) I made a ported over version of IBK built for Davinci Resolve as an ofx plugin. You can get it for free here: [IBKeymaster](https://dec18studios.com/color-grading-tools/ibkeymaster) It was originally brought from Nuke to Gaffer Tools by Jed Smith of Open DRT fame. What is IBKeymaster doing in Resolve that is better than Corridor Key! IBKeymaster is essentially already doing what the CK training pipeline does, just algorithmically in real-time instead of as a batch process with human oversight. What the CK Machine Learning (ML) Model Actually Adds The only thing the neural network genuinely gives you that algorithms can't: Semantic understanding: it "knows" that a wispy shape at the top of a head is probably hair, not noise. Our guided filter uses local statistics (variance, covariance) but has no concept of "hair" vs "screen wrinkle" Non-local context: the U-Net's receptive field spans the entire image. It can reason about "this shadow on the screen is consistent with the lighting direction from the key light." Our pipeline only sees local neighborhoods per kernel dispatch Everything else: the math of extracting alpha from color differences, cleaning plates, refining edges — we're already doing with dedicated, controllable, fast kernels. The Bottom Line The IBK System is an algorithmic version of the same pipeline that generates the CK ML training data. The CK ML model's only advantage is pattern recognition from training examples, and its disadvantages (black box, slow, no artistic control, training data dependency) are substantial. The IBK System is basically the training data pipeline, with the ability as an artist to tune every stage.