Rewrite Particle loops to be over objects, instead of integers

[erikvansebille]

Make the loops in Field and FieldSet that access Particles (in particular those two classes; possibly others too) independent of of the order - hence: migrate loops like

for i in range(numerical):
       new_value = array[i]

to either:

for obj in array:
      new_value = obj

or:

for i, obj in enumerate(array):
      new_array[i] = obj

This would simplify the integration of the linked list of nodes and the lists-of-arrays of particles (#862) later. This is because their indices are dynamic and could rearrange - a classic reason why to use iterators.

Thanks to @CKehl for bringing this up

[erikvansebille]

@CKehl, I've been looking into this Issue, but I don't see where these problematic loops are. Can you point them out to me?

[CKehl]

it's most likely in the 'Kernel', but those kind of loops exist in other classes too. Let me see where they are and collect them.

[CKehl]

Following lines are relevant: 'for p in range(pset.size)' (kernel.py; l. 291 and 372) list comprehension 'for i in range(pset.size)': (example_peninsula.py; ;. 150 and l. 187) other examples that have list comprehensions in the style 'for i in range(pset.size)', where either i or pset is named differently 'for i in range(self.size)' and 'for i in range(pset[_new].size)' (particleset.py; l. 265 and l. 687 and l. 776)

generally problem: if particle.id is not the same as it's position in the list (e.g. due to deletion), this is prone to failure.

[CKehl]

see #913 and follow-ups - issue resolved, progressively.