Add Dockerfile and build system for Jupyter image

Created docker/ directory with:
- Dockerfile.jupyter: Extends mcp-forge/python:3.12 with ipykernel 6.29.0
- build-images.sh: Build script with testing
- README.md: Complete documentation and usage instructions

Image Details:
- Base: mcp-forge/python:3.12 (python:3.12-slim)
- Size: 196 MB (~73 MB added for ipykernel + dependencies)
- Includes: ipykernel, jupyter-client, pyzmq, ipython, debugpy, etc.
- Version: 6.29.0 (matches host jupyter-client compatibility)

Build tested and verified:
✓ ipykernel imports correctly (6.29.0)
✓ kernel launcher works
✓ Image ready for use

Next: Update config and run integration tests

docker/Dockerfile.jupyter

@@ -0,0 +1,30 @@
+# Jupyter Kernel Image for MCP-Forge
+# 
+# This image provides a Python environment with ipykernel for stateful
+# execution via Jupyter protocol. It runs inside containers managed by
+# MCP-Forge and communicates with the host via ZMQ.
+#
+# Base: mcp-forge/python:3.12 (which is python:3.12-slim)
+# Adds: ipykernel for Jupyter kernel functionality
+
+FROM mcp-forge/python:3.12
+
+LABEL maintainer="MCP-Forge"
+LABEL description="Python 3.12 with ipykernel for stateful execution"
+LABEL version="1.0"
+
+# Install ipykernel
+# Using --no-cache-dir to keep image size small
+RUN pip install --no-cache-dir \
+    ipykernel==6.29.0 \
+    && python -m ipykernel install --user
+
+# Set working directory
+WORKDIR /workspace
+
+# The container will be run with:
+# - Host networking (for ZMQ communication)
+# - Mounted connection file
+# - Command: python -m ipykernel_launcher -f /tmp/kernel-{id}.json
+#
+# No ENTRYPOINT or CMD needed - command specified at runtime