Fix indexing with Slice

[jishnub]

Fixes #209

On master and Julia 1.0

julia> using OffsetArrays

julia> r = 1:2:10
1:2:9

julia> r[Base.Slice(Base.OneTo(3))]
ERROR: StackOverflowError:
Stacktrace:
 [1] getindex(::StepRange{Int64,Int64}, ::Base.Slice{Base.OneTo{Int64}}) at /home/jishnu/Dropbox/JuliaPackages/OffsetArrays.jl/src/OffsetArrays.jl:352 (repeats 80000 times)

After this PR:

julia> using OffsetArrays

julia> r = 1:2:10
1:2:9

julia> r[Base.Slice(Base.OneTo(3))]
1:2:5 with indices 1:3

[codecov]

Codecov Report

Merging #210 (36fbb3f) into master (23de436) will not change coverage.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##           master     #210   +/-   ##
=======================================
  Coverage   98.28%   98.28%           
=======================================
  Files           5        5           
  Lines         291      291           
=======================================
  Hits          286      286           
  Misses          5        5           

Impacted Files	Coverage Δ
src/OffsetArrays.jl	97.93% <100.00%> (ø)

---

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 23de436...36fbb3f. Read the comment docs.

[timholy]

I ended up having to try the tests to understand this, but now that I do the one thing I'd request is adding

r2 = IdentityUnitRange(2:10)

and

r2 = IdOffsetRange(Base.OneTo(9), 1)

to the tests. My reading is that it should still work, but given that no_offset_view performs "real work" in these cases that's a bit different from what's in the test coverage, I think we should check.

[jishnub]

I've added tests for IdOffsetRange(Base.OneTo(9), offset). Some tests for r2 = IdentityUnitRange(2:10) were already there, but now I've added some other r1 types to make the tests for r1[r2] more robust.

[rikhuijzer]

Thanks a lot for fixing this so quickly 😄 Could you also tag the patch/call JuliaRegistries register? CI is still failing at https://github.com/TuringLang/MCMCChains.jl.

[rikhuijzer]

@jishnub

[timholy]

JuliaRegistries/General#30890