`det_cond_forecast` interface is buggy

There is currently an implementation in the preprocessor that does not work correctly. Rework this.

changed milestone to %4.6

added bug documentation labels

As much as possible, we should keep the interface consistent with conditional_forecast (used for conditional forecasting on stochastic models).

This fix also requires the introduction of a preprocessor syntax for init_plan, basic_plan, and flip_plan.

changed title from det_cond_forecast interface is buggy to det_cond_forecast interface is not useable

If this is getting too complicated, maybe we can just drop the buggy interface.

I guess "requires" above is not correct. We can still fix this without adding interfaces for those other MATLAB/Octave functions.... But, perhaps we want to do away with the preprocessor interface for det_cond_forecast if we will not make one for init_plan, basic_plan, and flip_plan. I'm indifferent.

changed title from det_cond_forecast interface is not useable to det_cond_forecast interface is buggy

mentioned in commit dynare@71792cd6

aesthetic fixes: 42106f85

bug fix: 0cd59dbc

The interface was implemented in 139e3efa

It is not documented. Rather the MATLAB/Octave function is documented. See dynare#1688 (closed)

The interface is not used in the test files (the det_cond_forecast calls in tests/conditional_forecasts/5/fs2000_cal.mod are MATLAB/Octave statements).

If only one symbol is passed to det_cond_forecast there is an indexing error (see line 778 of ComputingTasks.cc).

Decisions:

1. Is this interface used? Should we keep it?
2. If we keep it, should it be documented, or was a decision made to leave it undocumented? If it should be documented, the test file should be modified to use it.
3. If we keep it, does it make sense to also create an interface for init_plan, basic_plan, and flip_plan as they are potentially used together? It seems awkward to have an interface for one but to leave the others as MATLAB/Octave statements.

added decision label

assigned to @MichelJuillard and unassigned @houtanb

I want to have a look at it because there are other issues related to conditional forecast

Just to make it explicit, fixing the interface for det_cond_forecast, init_plan, basic_plan, and flip_plan will most likely also solve the minor issue in dynare#1688 (closed)

mentioned in commit e137b2df

mentioned in commit dynare@3847df21

The unfinished interface for det_cond_forecast belongs to an ongoing work on linear block decomposition done by @FerhatMihoubi (see also dynare!1626 (closed)).

I have removed this interface in the 4.6 branch, since this work is not yet finished.

mentioned in commit 0cd59dbc