Support values::AbstractDict

[timholy]

This adds support for looking up values in an AbstractDict, not just an AbstractVector. The main intended consumer is ImageSegmentation, which does not necessarily use adjacent values for its segment labels: https://github.com/JuliaImages/ImageSegmentation.jl/blob/55c1310dbf6158d6fe1ec9fd765213ac5377bfbb/src/core.jl#L13-L14

I've bumped this to 0.5.2, but what about just going to 1.0? Even this change is non-breaking, and I'm having a hard time picturing major ways we might break this package in the future.

[codecov]

Codecov Report

Merging #25 (8ee63e7) into master (3a821ba) will not change coverage.
The diff coverage is 100.00%.

@@            Coverage Diff            @@
##            master       #25   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files            1         1           
  Lines           48        56    +8     
=========================================
+ Hits            48        56    +8     

Impacted Files	Coverage Δ
src/IndirectArrays.jl	100.00% <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 3a821ba...8ee63e7. Read the comment docs.

[johnnychen94]

Nice. 1.0 looks good to me.

[johnnychen94]

BTW, do you think it's the scope of this package to also handle "(lazy) big thunks" of indices JuliaImages/ImageCore.jl#170 (comment)?

[timholy]

You mean specifically the chroma-subsampling? I've not put enough thought into it to say for certain, but it seems a bit different (it's a manipulation of the pixel index rather than the pixel value).

[timholy]

OK, I switched this to 1.0 and took advantage of the opportunity to drop Julia 0.7. I assume we want to keep 1.0?

I'll leave this open for a couple of days, since I know that percolating [compat] is annoying to everyone and we should give a chance for downstream users to comment. RegistryCompatTools flags a couple of packages (but not many): CC @Mattriks (Gadfly), @RobBlackwell (EchogramImages), @lorenzoh (FastAI), @brenhinkeller (StatGeochemBase).

[brenhinkeller]

Sounds cool to me!

[timholy]

I should have also CCed @tlnagy, but I have already tested TiffImages locally and I know it passes without issue (after bumping the [compat]).