calculate storage from Basin / profile

[visr]

Fixes #225

This removes the need to provide the storage for a profile. We can calculate it ourselves by integrating over A(h) to get A(S) and h(S).

First this updates the test models to a correct storage in the profile. Then we calculate the profile and get the same results. Then we remove the storage from the schemas and docs.

I hoped this would help with #80 but based on the plot I added there it doesn't seem to. This does not yet add all the validation and extrapolation for lookup tables as described in #279, that can be done separately.

[SouthEndMusic]

I'll try adding the proper interpolation.

[SouthEndMusic]

I suggest something like this for the interpolation:

function get_area_and_level(basin::Basin, state_idx::Int, storage::Float64)
   storage_discrete = basin.storage[state_idx]
   area_discrete = basin.storage[state_idx]
   level_discrete = basin.store[state_idx]
  
   # storage_idx: smallest index such that storage_discrete[storage_idx] >= storage
   storage_idx = searchsortedfirst(storage_discrete,storage)

   if storage_idx == 1 # If the lowest discrete_storage level 0, this can only happen if the storage is 0 since storage is never negative
      level = level_discrete[1]
      area = area_discrete[1]

   else
      if storage_idx == length(storage_discrete)+1
         # These values yield linear extrapolation of area(level) based on the last 2 values
         area_higher = area_discrete[end]
         area_lower = area_discrete[end-1]
         level_higher = level_discrete[end]
         level_lower = level_discrete[end-1]
         storage_lower = storage_discrete[end]
      else
         area_higher = area_discrete[storage_idx]
         area_lower = area_discrete[storage_idx-1]
         level_higher = level_discrete[storage_idx]
         level_lower = level_discrete[storage_idx-]
         storage_lower = storage_discrete[storage_idx-1]
        
      area_diff = area_higher - area_lower
      level_diff = level_higher - level_lower # This should never be 0, validate?

      if area_diff == 0
         # Constant area means linear interpolation of level
         area = area_lower

         storage_higher = storage_discrete[storage_idx]
         level = level_lower + level_diff*(storage-storage_lower)/(storage_higher-storage_lower)
         
      else
         area = sqrt(area_lower^2 + 2*(storage-storage_lower)*area_diff/level_diff)
         level = level_lower + level_diff*(area-area_lower)/area_diff
      end
   end

   return area, level
end

Here I assume that the discrete storage variables are normal vectors. Should we also require that the profile is given by at least 2 height values?