Proposal #101: Add GLM-5.3-Flash, remove Kimi-K2.6 and GLM-5.2-FP8 from PoC models

Add zai-org/GLM-5.3-Flash as a governance-approved model and PoC model. Remove moonshotai/Kimi-K2.6 and zai-org/GLM-5.2-FP8 from poc params.models. MiniMaxAI/MiniMax-M2.7 and deepseek-ai/DeepSeek-V4-Flash-0731 are not affected: their PoC parameters and weights remain unchanged. All other chain parameters remain unchanged. The GLM-5.3-Flash model and its parameters were proposed by the kaitaku.ai team, independently validated by vbgd0, and are now proposed jointly. Release inputs, measurements and the reasoning behind the thresholds are in gonka-ai/gonka 1734. GLM-5.3-Flash PoC parameters: seq len 1024, dist threshold 0.44, p mismatch 0.10, p value threshold 0.05, weight scale factor 0.62; validation threshold 0.951. The weight scale factor was calibrated so that a B200 host switching its PoC model to GLM-5.3-Flash is expected to gain approximately 7% more weight relative to its current weight. For other GPUs the optimal model does not change. The activation epoch (penalty start epoch) for GLM-5.3-Flash is 394.