clustering: when using -a the number of input streamlines printed in the `Clustering` part is not the one actually used, because it doesn't take into account of the non-connecting streamlines discarded during the split.

[ilariagabusi]

E.g.:
!! note also the typo in _T_hreshold !!

Dividing the streamlines into anatomical bundles                           
 WARNING  Atlas data type is 'float32'. It is recommended to use an integer
data type.                                                                 
   * Computing assignments 
   * Number of regions: 85
[ 01m00s ]                                                                 
Splitting the bundles into separate files                                  
 WARNING  302896 streamlines were not assigned to any bundle               
[ 19s ]                                                                    
Clustering                                                                 
   * Number of input streamlines: 3000000
   * Clustering hreshold: 2.0
   * Number of points: 12
   * Computing workload for parallel clustering 
   * Parallel bundles clustering 
   * Number of computed centroids: 669671
[ 36s ]