Merge branch 'global_memory_barriers' into 'master'

doc/tutorial.rst

loopy/codegen/control.py

loopy/kernel/creation.py

loopy/kernel/instruction.py

loopy/kernel/tools.py

loopy/schedule/__init__.py

loopy/schedule/device_mapping.py

loopy/statistics.py

loopy/target/__init__.py

loopy/target/cuda.py

loopy/target/ispc.py

loopy/target/opencl.py

loopy/transform/add_barrier.py

loopy/transform/precompute.py

loopy/transform/save.py

loopy/version.py

test/test_loopy.py