Add partial isequal for VersionNumber

[IanButterworth]

Instead of

julia> v1 = v"1.6"; v2 = v"1.6.3";

julia> v1.major == v2.major && v1.minor == v2.minor
true

this provides

julia> isequal(v"1.6", v"1.6.3", downto = :minor)
true

julia> isequal(v"1.6", v"1.6.3", downto = :patch)
false

julia> isequal(v"1.6", v"1.6.3")
false

[IanButterworth]

@StefanKarpinski I'd appreciate a review. In particular I didn't want to add something that would be considered bad for semver best practice

[IanButterworth]

Given no red flags have been raised I'll go ahead

[mcabbott]

I am far from an expert on semver, but this seems a bit weird. I hoped a real expert would chime in.

What seems odd to me is that this is symmetric. If you have x = v"1.6", for what purpose would you want to accept both v"1.5" and v"1.7" equally? I would have thought the more common thing to want is an asymmetric check, which would accept only things on one side, thus v"1.7"but not v"2.1" (or v"1.5" but not v"0.9"), which > doesn't do.

[KristofferC]

Sorry for post merge review, but I find it kind of strange to add specific keyword arguments to an extended generic function. You can only use the keyword argument if you know the specific type but then why should the generic function be extended at all. The generic function should be documentable and there is no way to document the keyword arguments anymore since they are depending on concrete implementations.

This could perhaps be its own function that doesn't extend anything else.

[IanButterworth]

Ok. Seems I was too eager here for a feature that's only a slight convenience. I'll revert.

[StefanKarpinski]

Didn't see this, but I'm glad it's reverted. In general, I object to isequal ever acting in a non-transitive way, which this patch would have implemented. Note that you can do this with the thismajor, thisminor, thispatch functions, e.g. thisminor(v1) == thisminor(v2). We could export those functions potentially.