fix indexing of waveform files with duplicated traces

before, indexing failed if a file contained at least one duplicated trace, i.e.
two traces with exactly the same NSLC and start/endtime. this happens sometimes
in real life during data retrieval glitches (e.g. two slarchive threads running
at same time by mistake).

the database layout has to be changed, so that the Trace's position in the
indexed file (field "pos") is included in the unique constraint of the waveform
trace table.

src/jane/waveforms/models.py

@@ -102,8 +102,8 @@ def __str__(self):
 
     class Meta:
         ordering = ['-timerange', 'network', 'station', 'location', 'channel']
-        unique_together = ['file', 'network', 'station', 'location', 'channel',
-                           'timerange']
+        unique_together = ['file', 'pos', 'network', 'station', 'location',
+                           'channel', 'timerange']
 
     def timed_preview_trace(self):
         num_samples = (len(self.preview_trace) - 1)