Origin-center grids by default; fix a type conversion; update tests

[evanfields]

• Update the tests to ensure some basic behavior. The test coverage is still quite minimal, but nonetheless expanded relative to master. And I think all the behavior tested for should be uncontroversial.
• Set the default x and y offset for center to be 0, not 1. This means that the zero index hexagon (e.g. HexagonCubic(0, 0, 0)) will be centered at the xy origin. This should fix both inconsistency in cube_round #13 and hexbin misalignment Gadfly.jl#1497 .
• Fix axial to cubic conversion. The current master has

function convert(::Type{HexagonCubic}, hex::HexagonAxial)
    HexagonCubic(hex.q, hex.r, -hex.q - hex.r)
end

compare to Amit Patel's reference:

function axial_to_cube(hex):
    var x = hex.q
    var z = hex.r
    var y = -x-z
    return Cube(x, y, z)

i.e. the -q-r term should become the cubic y (second argument), but current master puts this term in cubic z.

[evanfields]

As far as I can tell, this section is no longer necessary with center's default offset at zero, not 1. The redblobgames references doesn't describe this neighbor check code. Seems like this latter loop was previously necessary because all hexagons were shifted by (-1,-1) in Cartesian space, so the shifted hexagon that contains a Cartesian point might not be the unshifted hexagon that contains that point.

With this shift behavior removed, nearest_cubic_hexagon is already returning the correct (0-centered) hexagon for the input point.

[evanfields]

@bjarthur apologies for the pestering tag but looks like you have merge privileges here and this would fix the downstream Gadfly bug. Let me know if you'd like any changes!