docs: Field Notes article, model card, charts; add traces publisher

Default extraction to the fine-tuned MiniCPM-V Hub checkpoint.

Remove dead chat code and obsolete planning docs.

feat: medical-reasoning LoRA fine-tune (medical-o1) — pin ms-swift, background spawn

feat: medical-reasoning LoRA fine-tune (medical-o1) for the interpretation layer

feat: real-report labeling helper + real-data mix-in for fine-tune

feat: gentler fine-tune + diversified synth data + interpretation engine + before/after eval

feat: working Modal LoRA fine-tune + llama.cpp-on-ZeroGPU backend

fix(train): use MiniCPM-V 4.6 model id and official mmproj

feat(train): Modal MiniCPM-V LoRA fine-tune + GGUF conversion scripts

feat(train): synthetic lab-report generator + SFT conversion