Refactor coordEach utilising geomEach internally // alternative approach to #57

[lukas-h]

As described in my comment here:
#57 (comment)

I have built an implementation where I could get rid of many of the helper functions by implementing it internally with geomEach. I just wanted to test if that is possible, but seems to work. You can find it here:

in response to @tobrun's comment here: #57 (comment)

This PR proposes the refactored code from #50 to match geomEach setup.
[...]
Would you take a different approach?

what do you think?

[tobrun]

Looks great to me, might be good if we can run the benchmarks for both the old branch as this and see if there is a measurable difference

[lukas-h]

I will try to do it!

[lukas-h]

Branch lukas-idea-coordeach-refactor:

 coordEach
  ✓ geometry (0 us)
    Elapsed:    2 s
    Rounds:     2274999
    Iterations: 10
  ✓ feature (0 us)
    Elapsed:    2 s
    Rounds:     2581970
    Iterations: 10
  ✓ geometry collection (200 us)
    Elapsed:    2 s
    Rounds:     9994
    Iterations: 10
  ✓ feature collection (218 us)
    Elapsed:    2 s
    Rounds:     9137
    Iterations: 10

Branch tvn-coordeach-refactor:

coordEach
  ✓ geometry (0 us)
    Elapsed:    2 s
    Rounds:     3546893
    Iterations: 10
  ✓ feature (0 us)
    Elapsed:    2 s
    Rounds:     3860529
    Iterations: 10
  ✓ geometry collection (0 us)
    Elapsed:    2 s
    Rounds:     7164767
    Iterations: 10
  ✓ feature collection (102 us)
    Elapsed:    2 s
    Rounds:     19512
    Iterations: 10

Branch main:

 coordEach
  ✓ geometry (0 us)
    Elapsed:    2 s
    Rounds:     3125071
    Iterations: 10
  ✓ feature (0 us)
    Elapsed:    2 s
    Rounds:     3323376
    Iterations: 10
  ✓ geometry collection (0 us)
    Elapsed:    2 s
    Rounds:     3248169
    Iterations: 10
  ✓ feature collection (223 us)
    Elapsed:    2 s
    Rounds:     8942
    Iterations: 10

@tobrun

[lukas-h]

At the moment, it looks like, your version @tobrun is the fastest.

But I suspect, that you are skipping GeometryCollection altogether, which causes the 0 us here. I may be wrong (#57 (comment))

[tobrun]

Let's go with this version!