revert to 8-step Q5_0 GGUF (4-step OOMs during build conversion)

fix: use q5_0 quantization type (q5_k_m not supported by sd.cpp)

4-step distill: download BF16, convert to Q5_K_M GGUF at build time

remove --schedule beta (not supported by sd.cpp)

Z-Anime 6B CPU: distill 8-step Q5_0, Qwen3-4B Q8_0, euler_a, beta schedule

initial commit