Check whether input landmarks are double in matlab-fitting

Matlab's default precision is double, that's why we only convert MatrixXd for now. Added a note highlighting this.

matlab/+eos/+fitting/fit_shape_and_pose.m

@@ -15,7 +15,7 @@ function [mesh, rendering_parameters] = fit_shape_and_pose(morphable_model, ...
 %   landmarks must be a 68 x 2 matrix with ibug landmarks, in the order
 %   from 1 to 68.
 %
-%   Default values for some of the parameters:: num_iterations = 5,
+%   Default values for some of the parameters: num_iterations = 5,
 %   num_shape_coefficients_to_fit = all (-1), and lambda = 30.0.
 %
 %   Please see the C++ documentation for the description of the parameters:
@@ -26,6 +26,10 @@ function [mesh, rendering_parameters] = fit_shape_and_pose(morphable_model, ...
 %   contour_landmarks and model_contour as *filenames* to the respective
 %   files in the eos/share/ directory, and not the objects directly.
 
+if(~isa(landmarks,'double'))
+    error('Please specify the landmarks as type double.');
+end
+
 % We'll use default values to the following arguments, if they're not
 % provided:
 if (~exist('edge_topology', 'var')), edge_topology = '../share/sfm_3448_edge_topology.json'; end

matlab/include/mexplus_eigen.hpp

@@ -33,6 +33,8 @@ namespace mexplus {
 /**
  * @brief Define a template specialisation for Eigen::MatrixXd for ... .
  *
+ * The default precision in Matlab is double, so for now we only define conversions for MatrixXd.
+ *
  * Todo: Documentation.
  */
 template<>