[ci skip] Two level scan now can be done in parallel.

loopy/kernel/__init__.py

loopy/kernel/tools.py

loopy/preprocess.py

loopy/schedule/tools.py

loopy/transform/instruction.py

loopy/transform/reduction.py

loopy/transform/save.py

test/test_loopy.py

test/test_scan.py

test/test_transform.py