more efficient unary_union in h3fy

[codecov-io]

Codecov Report

Merging #126 (179fe90) into master (399ea20) will increase coverage by 0.20%. The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #126      +/-   ##
==========================================
+ Coverage   82.32%   82.52%   +0.20%     
==========================================
  Files          15       15              
  Lines         758      767       +9     
==========================================
+ Hits          624      633       +9     
  Misses        134      134              

Impacted Files	Coverage Δ
tobler/tests/test_utils.py	100.00% <100.00%> (ø)
tobler/util/util.py	89.47% <100.00%> (+0.14%)	:arrow_up:

---

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 399ea20...e401c16. Read the comment docs.

[martinfleis]

This does not help in case of multipolygons. We do unary_union, then fail -> except: and do it again.

I'd do unary_union, save it separately and pass it directly to _to_hex and that loop in case of multipolygon instead of passing gdf.

[knaaptime]

gotcha. the new strategy should handle it this way

[martinfleis]

Hey @knaaptime, I have pushed a bit of a cleanup. There's no need to go via geopandas, explode and such in this. Feel free to revert if you don't like this way but it seems to be cleaner and more straightforward to my eyes like this.

[knaaptime]

excellent, thank you! I was just in the middle of doing a portion of this because my crs handling was ridiculous. Looking at the code for too long and jumping through too many hoops :)

[knaaptime]

@martinfleis i had to move the clip up above the reprojection otherwise the gdfs are misaligned, otherwise i think this is set