[edf] fix efficient array handling

Author: sprenger

neo/rawio/edfrawio.py

@@ -197,12 +197,13 @@ def _get_analogsignal_chunk(self, block_index, seg_index, i_start, i_stop,
             self.edf_reader.read_digital_signal(channel_idx, i_start, n, buffer)
             data.append(buffer)
 
+        # downgrade to int16 as this is what is used in the edf file format
+        # use fortran (column major) order to be more efficient after transposal
+        data = np.asarray(data, dtype=np.int16, order='F')
+
         # use dimensions (time, channel)
         data = data.T
 
-        # downgrade to int16 as this is what is used in the edf file format
-        data = np.asarray(data, dtype=np.int16)
-
         return data
 
     def _spike_count(self, block_index, seg_index, spike_channel_index):