Fixing bugs in unit aggregation and the selection of unit efficiency approximations. It is now easier to skip some efficiency approximation levels.

Issue #135

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

Previously forecast offset `df_scenario` was computed only when number of scenarios was
greater than one. Long-term scenario data should use a special label in the `f` dimension
in all cases.

Set `t_realized(t)` has the realized time steps in the simulation (excluding the starting point and any extra horizon).
This is included in file *results.gdx* as symbol `t`.

Closes #81

This is currently expressed by 'energyStoredPerUnitOfState'.

This way the explanations can be exported programmatically.

Adding two new parameters under param_policy: reserve_activation_duration and reserve_reactivation_time. The parameters are now included in q_stateUpwardLimit, q_stateDownwardLimit and q_boundStateMaxDiff. Default value for both parameters is 1 (hour) and reserve_reactivation_time should be greater than or equal to reserve_activation_duration.

Issue #129.

Issue #137

Converting from 1.2 to 1.3: include p_gn(grid,node,'nodeBalance')=1 for grid-nodes that don't have values in ts_priceChange.

Changing the interpretation of unitUnitEffLevel. The model was reading the input data so that 'effLevel to start aggregation' meant the start of the next level (e.g. using level3 resulted in the last time step of level3 in the internal parameter lastStepNotAggregated - in effect pointing to the start of Level4). Now this is read so that level3 means the start of level3

Adding domain checking for a few parameters and sets. Removing p_gnPolicy and p_groupPolicy3D, and adding p_groupPolicyEmission and p_groupPolicyUnit. Affects input files. Issue #121.

Updating the right hand side of q_unitEqualityConstraint and q_unitGreaterThanConstraint (issue #124). p_unitConstraint should take into account the size of one sub-unit as it will be multiplied by the number of online sub-units (or 1 if the unit does not have an online variable).

new parameters added to p_groupPolicy to allow for selection of static and/or dynamic inertia constraints

`p_storageValue` is the constant form parameter, while `ts_storageValue` is the corresponding timeseries. `ts_storageValue_` is also added to handle time step aggregation and data circulation of `ts_storageValue` in the objective function. A new flag in `p_gn` called `storageValueUseTimeSeries` controls whether the constant `p_storageValue` or the timeseries `ts_storageValue` is used.

WIP: I have no clue what the file `1e_scenmChanges.gms` is supposed to do, so no changes have been made there. However, since the input data is affected by these changes, I assume that this file will need to be changed as well for this fix to work properly.