Custom Features, use exogeneous data

As explain in issues/176, users can access to a list of bands by sensors to compute their own spectral features. However, we can imagine than externals data (ie : time series of masks) can be needed to compute a temporal feature (max NDVI considering clouds...).

I see two kind of external data :

• masks from sensors (which are already computed by iota2)

    def get_soi(self):
        """
        compute the Soil Composition Index
        """
        # for instance
        validity_vector = self.get_Sentinel2_clouds_series()
        ...
        coef = (self.get_Sentinel2_B11() - self.get_Sentinel2_B8()) / (
        self.get_Sentinel2_B11() + self.get_Sentinel2_B8())
        labels = [f"soi_{i+1}" for i in range(coef.shape[2])]
        return coef, labels

• a stack of features already prepared by users, they can access in their functions.

  def get_soi(self):
      """
      compute the Soil Composition Index
      """
      # for instance
      inner_vect_info = self.get_user_data_series()
      ...
      coef = (self.get_Sentinel2_B11() - self.get_Sentinel2_B8()) / (
      self.get_Sentinel2_B11() + self.get_Sentinel2_B8())
      labels = [f"soi_{i+1}" for i in range(coef.shape[2])]
      return coef, labels