Add capability to reset the "state" of the mass balance model

docs/whats-new.rst

@@ -21,6 +21,13 @@ Enhancements
   ``dis_from_border``, ...) to define this ranking, providing greater flexibility
   and control over how the melting sequence is visualized (:pull:`1746`).
   By `Patrick Schmitt <https://github.com/pat-schmitt>`_
+- Added "reset_state()" function to ``MassBalanceModel`` to signal any state-dependent 
+  mass balance model to "reset its state" at the start of the period. This function
+  does nothing in the parent class and would only be implemented by a state-ful
+  mass balance model. Done mainly to address that ``apparent_mb_from_any_mb()`` calls 
+  ``get_specific_mb()``  twice, leading to nonzero terminal flux (potentially) with a 
+  state-dependent mass balance model. `apparent_mb_from_any_mb()` also changed 
+  accordingly. 
 
 Bug fixes
 ~~~~~~~~~

oggm/core/massbalance.py

@@ -66,6 +66,14 @@ def __repr__(self):
                 summary += [nbform.format(k, v)]
         return '\n'.join(summary) + '\n'
 
+    def reset_state(self):
+        """Reset any internal state of the model.
+        This might not be needed by most models, but some models have an
+        internal state (e.g. a snow cover history) which can be reset
+        this way.
+        """
+        pass
+
     def get_monthly_mb(self, heights, year=None, fl_id=None, fls=None):
         """Monthly mass balance at given altitude(s) for a moment in time.
 
@@ -2087,7 +2095,7 @@ def to_minimize(ela_h):
 @entity_task(log, writes=['inversion_flowlines'])
 def apparent_mb_from_any_mb(gdir, mb_model=None,
                             mb_model_class=MonthlyTIModel,
-                            mb_years=None):
+                            mb_years=None, reset_state=False):
     """Compute apparent mb from an arbitrary mass balance profile.
 
     This searches for a mass balance residual to add to the mass balance
@@ -2110,6 +2118,10 @@ def apparent_mb_from_any_mb(gdir, mb_model=None,
         geodetic MB period, i.e. PARAMS['geodetic_mb_period'].
         It does not matter much for the final result, but it should be a
         period long enough to have a representative MB gradient.
+    reset_state : bool
+        if True, the mass balance model will be reset at each mass-balance year.
+        This is useful if the model has an internal state that might
+        affect the results.
     """
 
     # Do we have a calving glacier?
@@ -2141,6 +2153,9 @@ def to_minimize(residual_to_opt):
 
     residual = optimize.brentq(to_minimize, -1e5, 1e5)
 
+    if (reset_state):
+        mb_model.reset_state()
+
     # Reset flux
     for fl in fls:
         fl.reset_flux()