Chunkwise iteration over arrays.

[jeromekelleher]

Closes #398.

Thanks to @alimanfoo for giving me the formula for unit tests! The only tricky thing is the zero-d case, which I've mirrored numpy's behaviour for.

TODO:

• Add unit tests and/or doctests in docstrings
• Add docstrings and API docs for any new/modified user-facing classes and functions
• New/modified features documented in docs/tutorial.rst
• Changes documented in docs/release.rst
• Docs build locally (e.g., run tox -e docs)
• AppVeyor and Travis CI passes
• Test coverage is 100% (Coveralls passes)

[alimanfoo]

Thanks @jeromekelleher, looks good.

Travis PY27 is failing because in PY2 it's itertools.izip_longest and PY3 it's itertools.zip_longest. Suggest to add something to the zarr.compat module to alias izip_longest to zip_longest under PY2.

Travis PY37 is failing with:

py37 runtests: commands[4] | flake8 zarr
zarr/storage.py:1452:25: E117 over-indented

...that's nothing to do with this PR, although I have no idea how that's crept in, but easiest thing to do would be to fix in this PR by dedenting that line.

Also please add an entry in the release notes (docs/release.rst).

[alimanfoo]

Suggested change

	chunk = self[j: j + chunk_size][:]
	chunk = self[j: j + chunk_size]

Additional slice shouldn't be necessary.

[jeromekelleher]

I thought you would need to take an extra copy to be sure the data was kept in a local numpy array. Changed now.

[alimanfoo]

Thanks, yeah just the single slice [j: j + chunk_size] is enough to bring data into a local numpy array. Adding the extra [:] is harmless but not necessary.

[jeromekelleher]

Thanks @alimanfoo, I've implemented those. Love the changelog format btw, I think I'll be copying that!

[alimanfoo]

LGTM, thanks!