feat(interpreters): implement ipynb streaming parser with 97.9% memory reduction

[patchmemory]

Summary

Refactors Jupyter notebook interpreter to use pure streaming parsing via ijson, achieving 97.9% memory reduction for large notebooks (far exceeds 40% target).

Changes

• Removed full-load fallbacks: Eliminated 86 lines of code that defeated streaming purpose
• Made ijson required dependency: Added to pyproject.toml and requirements.txt
• Fixed cell counting: Parser now counts all cells accurately while limiting content sampling
• Version bump: 0.2.0 → 0.3.0
• Added comprehensive tests: 4 new memory profiling tests validate efficiency
• Documentation: Created tutorial explaining the optimization

Memory Impact

For a 3.6MB notebook (1,000 cells):

• Before: ~8MB peak memory
• After: ~165KB peak memory
• Reduction: 97.9%

Test Results

✅ All 6 ipynb interpreter tests pass
✅ Memory reduction test validates >=40% target (achieved 97.9%)
✅ Cell counting accuracy verified with 1,500 cell notebook
✅ Small notebook efficiency test passes (< 1MB peak)

Files Changed

• pyproject.toml - Added ijson dependency
• requirements.txt - Added ijson dependency
• scidk/interpreters/ipynb_interpreter.py - Refactored to pure streaming
• tests/test_ipynb_interpreter.py - Added memory profiling tests
• docs/ipynb-streaming-optimization.md - Tutorial documentation

Migration Notes

Zero API changes required. Existing code continues to work unchanged.

Resolves task:interpreters/refactor/ipynb-streaming

🤖 Generated with Claude Code