Functionality for slicing with unit ranges that preserves block information

Indexing with UnitRange does not preserve the blocking of blocked arrays:

julia> using BlockArrays

julia> r = blockedrange([2, 4])
2-blocked 6-element BlockedUnitRange{Int64, Vector{Int64}}:
 1
 2
 ─
 3
 4
 5
 6

julia> r[2:4]
2:4

and:

julia> a = BlockArray(randn(4, 4), [2, 2], [2, 2])
2×2-blocked 4×4 BlockMatrix{Float64}:
 0.372344   -0.735109  │   0.316133  -0.935035
 2.37641     1.30943   │  -0.931113   0.339143
 ──────────────────────┼──────────────────────
 0.422144    0.839572  │  -1.4721    -0.74205 
 0.0436636  -0.722946  │   0.799941  -0.192523

julia> a[2:3, 2:3]
2×2 Matrix{Float64}:
 1.30943   -0.931113
 0.839572  -1.4721

It would be nice to have convenient functionality for indexing with UnitRange that does preserve blocking.

Start by reproducing the UnitRange examples for blockedrange and BlockArray in Julia, then inspect the existing indexing behavior for BlockedUnitRange and BlockArray. Define how sliced ranges and arrays should retain block information; done means both examples return blocked results with appropriate block boundaries.