fix: harden Hugging Face deployment

feat: add Rule 11 (Mandatory Git Push), DirectML/OpenVINO EP ONNX, 1-Click Presets, Organ-Based Toggles, AI Quality Score & Multi-Scale Adaptive Sharpening

feat: add Reinhard Auto Skin Tone Alignment for seamless face color matching

feat: integrate Wink Quality Engine, skin grain retention, eye/lip sparkle, and update Obsidian vault

perf: implement lightweight pipeline optimizations and lazy memory release

fix: remove static quantization from Dockerfile build and add fail-safe ONNX loader

fix(hf): integrate Static INT8 ONNX build into Dockerfile and fix thread state guard in app.py

feat: implement static INT8 ONNX quantization tool & UI model engine selector

fix: resolve all 9 codebase bugs (B1-B9) & add Rule 9 to AGENTS.md

fix: patch B2, B6, B10, B12 (ONNX tiling upscaler, start_time guard, cache split screen comparison, list css style)

fix: patch B3 and B4 (ONNX mutex lock and async batch processing queue)

fix: patch B1, B5, B7, B8, B9 (state guards, dead callback, face size tuple, dashboard loss, esc key)

fix: resolve streamlit thread safety issues and cache face helper to optimize cpu

fix: resolve ValueError and stride mismatch when comparing parsing and soft boundary masks

fix: resolve Hugging Face deployment crashes and dependency conflicts

feat: complete custom training pipeline bugfixes, dataset crawling, and validation

feat: optimize training parameters, fix pipeline sequential bug, update UI default fidelity, and configure HF Spaces auto-deployment in Docker mode

feat: add Enable Face Restoration toggle in Streamlit UI sidebar and pipeline.py to support bypassing face cropping and restoration when upscaling already sharp images

fix: replace bilinear resize with Lanczos fallback and implement high-res original face blending based on fidelity weight to prevent degradation of sharp input images

opt: quantize Real-ESRGAN to INT8, add OpenVINO EP support to pipeline, add interactive comparison slider to streamlit, and add blur filtering to face crop

feat: add Safebooru crawler and facexlib-based face crop/align tool, implement progress callback in streamlit app

Fix training crash on CPU, reduce val_freq, add OpenCV env vars

CPU optimizations and UI defaults

Optimize CPU performance: add INT8 quantization, set default lightweight detector, disable heavy upscaling by default

feat: optimize CPU performance and add agents rules

fix: upscale restored face crops to prevent blurriness in high-resolution downloads

feat: add Google Colab GPU notebook, ONNX export script, and pipeline ONNX optimization

feat: implement post-processing sharpening filter and deploy custom model weights (iter 350)

feat: integrate Real-ESRGAN background upscaling and face detection threshold

chore: configure repository for Hugging Face Spaces Streamlit deployment

feat: implement custom enhancement pipeline with adjustable soft blending