Add cohorts() function

[mcelrath]

Notes:
You were setting the genesis block (0) as its own parent, I removed this as it borks the cohort algorithm. It doesn't seem to affect your code otherwise.

I had to compute the children array, which is just a reversal of direction on the parents array. next_generation is a generalization of both that allows traversal of the graph from genesis to tip or backwards from tip to genesis.

int/nb.int32 differences don't seem to have any impact but I can force it to use np.int32 if you want.

I didn't use this cohorts() function anywhere, just providing its definition here so you can use it instead of a correction factor.

Use this like:

for c in cohorts(): print(c) # Starts from genesis
for c in cohorts(12999, True): print(c) # starts from tip

[mcelrath]

I tested this like:

from DAG_braidpool_simulator import *
# let it run and generate its graph. Close graph window.
for c in cohorts(): print(c)