Use higher precision in imrotate

[timholy]

Fixes #147
Fixes #145

[timholy]

Let me acknowledge that the broken test isn't ideal. nvm, I figured out a fix. Still, it's worth pointing out:

julia> mod2pi(2π)
6.283185307179586

julia> mod2pi(nextfloat(2π))
6.432490598706546e-16

so it cannot exactly hit 0, and hence the NaN pattern at the edge is different. But it's the only case from among all the angles where there is a difference. But both differ from 0 or 2π by less than eps(2π) so we can still figure out what was meant.

[codecov]

Codecov Report

Merging #148 (1ea801b) into master (4d89e5c) will increase coverage by 0.84%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #148      +/-   ##
==========================================
+ Coverage   89.06%   89.90%   +0.84%     
==========================================
  Files           8        8              
  Lines         192      208      +16     
==========================================
+ Hits          171      187      +16     
  Misses         21       21              

Impacted Files	Coverage Δ
src/warp.jl	100.00% <100.00%> (ø)
src/resizing.jl	100.00% <0.00%> (ø)
src/autorange.jl	93.33% <0.00%> (+0.65%)	⬆️
src/interpolations.jl	75.00% <0.00%> (+3.00%)	⬆️

---

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 4d89e5c...1ea801b. Read the comment docs.

[johnnychen94]

Nice! Thank you for putting your efforts into this issue.

[timholy]

Just trying to catch up on rather a long time of letting you carry the load!