fix: use native GGUF chat template instead of name-based inference (#188)

fix: raise default n_ctx from 4096 to 8192 and fix UTF-8 token boundary (#187)

Bump version to 1.9.0

Fix Gate 1: Remove GPU backends (not available on GitHub runners), keep vision

Add vision feature to Gate 1 (Linux Kitchen Sink binary)

fix: remove duplicate Linux x86_64 build to prevent artifact conflict

fix: remove GPU backends from GitHub runner builds - CPU only

fix: remove CUDA from GitHub runner builds (no CUDA toolkit available)

fix: disable git version check in llama.cpp CMake build

fix: patch shimmy-llama-cpp-sys-2 to use git dependency for CI builds

fix: initialize git submodules for llama.cpp CUDA build

docs: create shimmy vision sales pipeline consolidation guide

feat: add critical testing for GPU backend robustness, vision performance, and license pipeline

chore: add simple wait-for-build script

chore: add build monitoring script for private repo