Make indexing idempotent again

[jishnub]

This is a patchwork of fixes that address a couple of common cases encountered while indexing with offset axes.

Now

julia> a = OffsetVector(3:4, 10:11)
3:4 with indices 10:11

julia> ax = Base.IdentityUnitRange(10:11)
Base.IdentityUnitRange(10:11)

julia> a[ax]
3:4 with indices 10:11

julia> axes(a[ax]) == axes(ax)
true

julia> ax = OffsetArrays.IdOffsetRange(5:6, 5)
OffsetArrays.IdOffsetRange(10:11)

julia> a[ax]
3:4 with indices 6:7

julia> axes(a[ax]) == axes(ax)
true

This is not really a proper fix, as ideally we would want to return an OffsetVector when an OffsetRange is indexed using an arbitrary AbstractUnitRange. However that might be too breaking, so as of now I suppose this will have to do.

Fixes #155

[codecov]

Codecov Report

Merging #156 into master will increase coverage by 0.83%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #156      +/-   ##
==========================================
+ Coverage   94.02%   94.86%   +0.83%     
==========================================
  Files           2        2              
  Lines         251      253       +2     
==========================================
+ Hits          236      240       +4     
+ Misses         15       13       -2     

Impacted Files	Coverage Δ
src/OffsetArrays.jl	95.69% <100.00%> (+1.00%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update dc867d5...96aa2c7. Read the comment docs.