New coding for fish distribution

[yunneshin]

There's a need to allow fish to not, systematically and by default, re-distribute randomly when a new distribution map is in input.
What I suggest as a quick fix:
1- add a boolean to each map, stating whether the new distribution map is a shift of distribution or not.
If it is a shift, all the schools are re-distributed randomly in the new map (this is the current default version).
If not, it's not a shift (smooth transition to a slightly changed area; ie the majority of the cells of the new map is common to the previous map), then I suggest to re-distribute randomly only the schools which are outside the new map.

2- the second option which can be put in place later is to work on the input maps before running the model. Once we have all the maps ordered (by season, age) for a given species, we identify the cells which will disappear with the new map. For each of this cell, we make a correspondence to the closest cell. We end up with a matrix of correspondence by pair of maps. That we apply to schools during the runs. This way, we avoid re-calculating the distances at each time step, but we do it only once, before running the model. the matrix between two maps would be part of the input data

[mtraverstrolet]

I also add some though about fish distribution, specially when you have different spawning grounds leading to different nursery grounds (link between eggs, larvae and juveniles distribution) but one big adult distribution. Maybe we could add some coding (eg numbers) between maps in order to identify where you can move depending on where you come from. For instance in this drawing, juveniles from blue area 1 would go as adult to orange area 1, but juveniles from blue area2 would go both to orange areas 1 and 2... or something like that, to prevent to much mixing when there is a strong spatial structure in the system...