Add modular system package installation support

Implemented SYSTEM_PACKAGES environment variable for flexible system library installation:

Key Features:
- Users specify packages in .env: SYSTEM_PACKAGES="libopenslide0 libgdal-dev"
- Packages installed during Docker build (secure, no runtime mounting)
- Automatic rebuild detection when packages change
- Documentation with common scientific computing packages

Technical Implementation:
- ARG SYSTEM_PACKAGES in Dockerfile with conditional apt-get install
- Build argument passed from claude-docker.sh script
- Clean package management with cache cleanup

Benefits:
- Secure: No system directory mounting required
- Flexible: Users choose exactly what they need
- Documented: Common packages listed in README
- Clean: Packages baked into image, not mounted at runtime

Tested successfully with OpenSlide: conda run -n prism python -c "import openslide"

.env.example

@@ -13,6 +13,16 @@ TWILIO_TO_NUMBER=+0987654321
 # Optional: Custom conda installation (for academic/lab environments)
 # Example: CONDA_PREFIX=/vol/biomedic3/username/miniconda3
 CONDA_PREFIX=
+
 # Optional: Additional conda directories (space-separated list)
+# Directories are mounted to the same path inside the container
+# Automatic detection:
+#   - Paths with "*env*" are added to CONDA_ENVS_DIRS (for environments)
+#   - Paths with "*pkg*" are added to CONDA_PKGS_DIRS (for package cache)
 # Example: CONDA_EXTRA_DIRS="/vol/biomedic3/username/.conda/envs /vol/biomedic3/username/conda_envs /vol/biomedic3/username/.conda/pkgs /vol/biomedic3/username/conda_pkgs /homes/username/.conda/envs"
-CONDA_EXTRA_DIRS=
+CONDA_EXTRA_DIRS=
+
+# Optional: System packages to install in Docker container (space-separated)
+# Common scientific packages: libopenslide0 libgdal-dev libproj-dev libopencv-dev
+# Example: SYSTEM_PACKAGES="libopenslide0 libgdal-dev"
+SYSTEM_PACKAGES=