use BitArray for burning buffer

[rafaqz]

@alex-s-gardner you could try this branch on your huge array without threaded=false

It should use 1/8th of the memory.

[alex-s-gardner]

This branch speeds things up considerably but for the large use case threaded=false is still much faster

Branch = Rasters v0.8.4
threaded=true: 260.770344 seconds (1.24 M allocations: 336.461 GiB, 0.09% gc time, 6.75% compilation time)
threaded=false: 8.425360 seconds (876.21 k allocations: 10.768 GiB, 0.88% gc time)

Branch = use_BitArray
threaded=true: 30.647470 seconds (932.37 k allocations: 42.203 GiB, 0.98% gc time)
threaded=false: 2.602775 seconds (913.17 k allocations: 1.491 GiB, 2.14% gc time)

[rafaqz]

Ok thats way lower allocation at least.

Maybe the threads are running into each other somehow, there is a lock in there but mostly it can be avoided. I'll have to look into that.

[codecov-commenter]

Codecov Report

Merging #533 (6441f08) into main (ff90b8f) will increase coverage by 7.92%.
Report is 2 commits behind head on main.
The diff coverage is 61.53%.

@@            Coverage Diff             @@
##             main     #533      +/-   ##
==========================================
+ Coverage   73.06%   80.98%   +7.92%     
==========================================
  Files          59       57       -2     
  Lines        4633     4108     -525     
==========================================
- Hits         3385     3327      -58     
+ Misses       1248      781     -467     

Files	Coverage Δ
ext/RastersArchGDALExt/warp.jl	96.00% <100.00%> (ø)
src/Rasters.jl	100.00% <ø> (ø)
src/methods/burning/array_init.jl	92.30% <100.00%> (-0.38%)	⬇️
src/methods/burning/geometry.jl	68.57% <ø> (ø)
src/methods/burning/point.jl	51.35% <100.00%> (ø)
src/methods/coverage.jl	92.92% <100.00%> (ø)
src/methods/mask.jl	90.00% <100.00%> (+0.38%)	⬆️
src/methods/rasterize.jl	89.91% <100.00%> (-0.03%)	⬇️
src/plotrecipes.jl	83.17% <ø> (+9.91%)	⬆️
src/stack.jl	84.47% <100.00%> (+0.09%)	⬆️
... and 1 more

... and 1 file with indirect coverage changes

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more