change reshape to vstack of padded arrays

when creating np.array from ragged list (list of arrays of differnt
lengths) for calc.projections, a visibledeprecationwarning was being
  issued. stacking padded arrays solves

**WARNING** : is this correct from the physical point of view ?
in case the atoms are given different functions, I think the Bs
of equal indexes are being misaligned.

Author: mdforti

src/pyace/asecalc.py

@@ -147,7 +147,11 @@ def calculate(self, atoms=None, properties=['energy', 'forces', 'stress', 'energ
 
         self.energy, self.forces = np.array(self.ace.energy), np.array(self.ace.forces)
         nat = len(atoms)
-        self.projections = np.reshape(self.ace.projections, (nat, -1))
+        ragged_projections = self.ace.projections
+        maxlen = max([len(this_proj) for this_proj in ragged_projections])
+        padded_projections = [np.pad(this_proj, (0, maxlen-len(this_proj))) for this_proj in ragged_projections ]
+        self.projections = np.vstack(padded_projections)
+#        self.projections = np.reshape(self.ace.projections, (nat, -1))
 
         self.energies = np.array(self.ace.energies)