Fix a bug in long-term scenario smoothing

Parameter `msEnd` was not used with `tSolveFirst` causing wrong values
in later solves.

CHANGELOG.md

@@ -4,6 +4,7 @@ All notable changes to this project will be documented in this file.
 ## 1.1.5
 ### Fixed
 - Long-term scenario data when using only one scenario
+- Bug with scenario smooting which caused wrong values on later than first solve
 
 ## 1.1.4
 ### Fixed

inc/3c_inputsLoop.gms

@@ -457,10 +457,13 @@ $ifthen.autocorr defined p_autocorrelation
 
 // Do smoothing
 if(mSettings(mSolve, 'scenarios'),  // Only do smooting if using long-term scenarios
-    // Select the initial sample, first `t` not in it and `f` of the last `t` in it
+    // Select the initial sample, the last time in it (t_)
+    // and the forecast (f_) of the last simulated time step (t__) in it
     loop((ms_initial(mSolve, s_), t_, ft(f_, t__))
-        $[ord(t_) = msEnd(mSolve, s_) and mst_end(mSolve, s_, t__)
-          and (mf_realization(mSolve, f_) xor mf_central(mSolve, f_))],
+        $[ord(t_) = msEnd(mSolve, s_) + tSolveFirst - 1
+          and mst_end(mSolve, s_, t__)
+          and (mf_realization(mSolve, f_) xor mf_central(mSolve, f_))
+         ],
         $$batinclude 'inc/smoothing.gms' ts_influx
         $$batinclude 'inc/smoothing.gms' ts_cf
     );