all code/utils.py to core/* && evaluate_h3cells overhaul

[jGaboardi]

This is a hefty PR that:

• moves all the functionality within code/utils.py to core/*
  • DRYs & enhances robustness
  • hex creation --> core.utils.py
  • stat analysis, etc. --> core.stats.py
  • plotting --> core.viz.h3_hex.py
• overhauls evaluate_h3cells functionality within code/
  • @anastassiavybornova one "major" change is that Hex cells can be isolated and plotted by hex_id -- more robust than index location
• tests for everything
• updates README
• xref update notebooks in code/ as functionality is added to core/ #63
• resolves add Ana's util func to __all__ & migrate to core.util.py #62
• resolves (slowly) emulate functionality in code --> core + test #60

[martinfleis]

I will let @anastassiavybornova review this one as it mostly touches her code.

[jGaboardi]

I will let @anastassiavybornova review this one as it mostly touches her code.

LOL Okay lazy bones.

[jGaboardi]

@martinfleis I think we'll actually want your eyes on a couple things for performance that @anastassiavybornova had mentioned when she made her original PR for this stuff. While I have reproduced her outputs (mostly) faithfully with some improved performance there is still the outstanding use of apply in cell 10 of evaluate_h3cells.ipynb as it relates to get_edge_stats() and get_edge_stats() in core/stats.py

[martinfleis]

It looks okay to me. Is there some performance issue? Does it take hours or anything?

[jGaboardi]

It looks okay to me. Is there some performance issue? Does it take hours or anything?

OK, cool. For Auckland that notebook cell takes ≈2 seconds to run. I wasn't sure if there was some easy adjustment or clever spatial join that could be done to cut that time. At any rate, I think good enough for now and we can revisit if it becomes a problem with larger AOIs.

[martinfleis]

If it took 20 minutes I would hesitate if it is worth any further attention. At 2s, you spend more reading this comment than what would you save in the end.

[anastassiavybornova]

Looks good to me, thanks James for the housekeeping as always <3 and yes sure, agree for the performance, if there's no palpable issue let's just leave the .apply (well, now it's .map) inside the stats functions