feat: initial dflash-server docker packaging

Multi-stage CUDA build of the native dflash_server from
Luce-Org/lucebox-hub (pinned at 42f36f1). Models are not baked
into the image; mount /models at runtime.

- Dockerfile: nvidia/cuda:12.6.0 devel -> runtime, CUDA_ARCH build-arg
  (default sm_86), libcuda.so.1 stub symlink + -rpath-link fix
- docker-compose.yml: reference service with ./models:/models:ro
- Makefile: submodules / doctor / build / run / shell / up-down-logs /
  push / clean. push targets gitea.va.reichard.io/evan
- README + .dockerignore + .gitignore

.dockerignore

@@ -0,0 +1,35 @@
+.git/
+.gitmodules
+.gitignore
+.dockerignore
+README.md
+DOCKER.md
+Makefile
+docker-compose.yml
+Dockerfile
+models/
+
+lucebox-hub/.git/
+lucebox-hub/dflash/build/
+lucebox-hub/dflash/build-*/
+lucebox-hub/dflash/.cache/
+lucebox-hub/dflash/.venv/
+lucebox-hub/dflash/**/__pycache__/
+lucebox-hub/dflash/**/*.pyc
+
+lucebox-hub/dflash/models/
+lucebox-hub/dflash/**/*.gguf
+lucebox-hub/dflash/**/*.safetensors
+lucebox-hub/dflash/**/*.bin
+
+lucebox-hub/dflash/docs/
+lucebox-hub/dflash/eval/
+lucebox-hub/dflash/tests/
+lucebox-hub/dflash/test/
+lucebox-hub/dflash/examples/
+lucebox-hub/dflash/RESULTS.md
+lucebox-hub/dflash/DEVELOPER.md
+lucebox-hub/dflash/CODEX.md
+lucebox-hub/dflash/*.gif
+lucebox-hub/dflash/hero*.png
+lucebox-hub/dflash/demo*.png