tsio/insert: storing the manual input like that is not really useful

Dominant use case is to receive things from excel, which
means the manual edition is actually embedded within
mostly "automatic" data. Very different from a carefully
crafted insert ...

So we'll keep the habit of computing the manual insertions
by diffing automatic against the base (synthetic).

test/test_supervision.py

+from pathlib import Path
 from datetime import datetime
 import pytest
 
@@ -420,14 +421,6 @@ def test_more_manual(engine, tsh):
 2015-01-05     True
 """, marker)
 
-    tshm = TimeSerie('tsh-manual')
-    man = tshm.get(engine, 'ts_exp1', _keep_nans=True)
-    assert_df("""
-2015-01-03   -3.0
-2015-01-04   -3.0
-2015-01-05    NaN
-""", man)
-
 
 def test_before_first_insertion(engine, tsh):
     tsh.insert(engine, genserie(datetime(2010, 1, 1), 'D', 11), 'ts_shtroumpf', 'test')

tshistory_supervision/schema.py

@@ -4,10 +4,8 @@ from tshistory.schema import init as tshinit, reset as tshreset
 def init(engine, meta, namespace='tsh'):
     tshinit(engine, meta, namespace)
     tshinit(engine, meta, '{}-automatic'.format(namespace))
-    tshinit(engine, meta, '{}-manual'.format(namespace))
 
 
 def reset(engine, namespace='tsh'):
     tshreset(engine, namespace)
     tshreset(engine, '{}-automatic'.format(namespace))
-    tshreset(engine, '{}-manual'.format(namespace))

tshistory_supervision/tsio.py

@@ -39,36 +39,27 @@ class TimeSerie(BaseTS):
       value.
 
     We can explain the workflow like with a traditional DVCS graph,
-    with three branches: "automatic", "manual" and "synthetic".
+    with two branches: "automatic" and "synthetic".
 
     All automatic fetches go into the automatic branch (and thus are
     diffed against each other).
 
     The synthetic series receive all the non-empty differences
-    resulting from inserting to the automatic series, and also
-    all the manual entries.
+    resulting from inserting to the automatic series, and also all the
+    manual entries.
+
+    The manual editions are computed as a diffs between synthetic and
+    automatic series.
 
-    The manual editions can be computed as a diff between synthetic
-    and automatic series, but for convenience we also store them
-    explicitly.
     """
 
     def __init__(self, *a, **kw):
         super().__init__(*a, **kw)
-        self.manual_store = BaseTS(namespace='{}-manual'.format(self.namespace))
         self.auto_store = BaseTS(namespace='{}-automatic'.format(self.namespace))
 
     def insert(self, cn, ts, name, author, _insertion_date=None, manual=False):
         if manual:
-            # insert & compute diff over synthetic
-            diff = self.manual_store.insert(
-                cn, ts, name, author, _insertion_date=_insertion_date
-            )
-
-            if diff is None:
-                return
             diff = ts
-
         else:
             # insert & compute diff over automatic
             diff = self.auto_store.insert(