Accept a num_elements_provider in FieldArray to specify a size in elements instead of bytes.

[rcfox]

After some thought, I was able to come up with a way to resolve #34 myself.

[rcfox]

Not sure what happened with the coverage build, but I don't think it's my fault.

coverage runtests: commands[1] | coveralls
Traceback (most recent call last):
  File ".tox/coverage/bin/coveralls", line 11, in <module>
    sys.exit(wear())
  File "/home/travis/build/digidotcom/python-suitcase/.tox/coverage/lib/python3.5/site-packages/coveralls/__init__.py", line 79, in wear
    from coveralls.control import coveralls
  File "/home/travis/build/digidotcom/python-suitcase/.tox/coverage/lib/python3.5/site-packages/coveralls/control.py", line 1, in <module>
    from coverage.control import Coverage
ImportError: cannot import name 'Coverage'
ERROR: InvocationError: '/home/travis/build/digidotcom/python-suitcase/.tox/coverage/bin/coveralls'

[mikewadsten]

I'd recommend using %s and %d, respectively, in this format string. Using %r will add single quotes which could be confusing.

Edit: Ignore.

[posborne]

I actually prefer %r in this case as %s formatting will not show some characters (binary non-printable). In addition, things like newlines will just be included and output when printed which can be very confusing.

[posborne]

@rcfox I can look at the coverage stuff and will look at this PR in the next few days (hopefully) once I get back from traveling.

[mikewadsten]

@posborne You're absolutely right on the %r stuff. I had it in my head that "%r %r" % (None, 1) yields the string 'None' '1', when that's obviously not the case.

[posborne]

Build failure should be fixed by #41 and I think this is a pretty valuable change to have in. I'm going to go ahead and include it.

Thanks and sorry for the long delay!