Issue #32: Run-up and shutdown trajectories overhauled to permit SOS2 ramping on the last/first run-up/shutdown intervals, as well as include main fuel consumption during run-up/shutdown. However, this formulation still fails to consider changing interval lengths.

Set `longtermSamples` was erroneously cleared too late

Issue #55: periodicLoop is now executed before inputsLoop, and the time series interval aggregation is moved to the end of inputsLoop.

-Removing the unitFail set (unit_fail set is directly read from the input GDX)
-Deleting N-1 param_policy (Infeed2Cover can handle its task)

Updating issue #47. It should now be possible to use same reserve for multiple categories. So, a unit can have a multiplier in the new table p_nuRes2Res indicating the ratio at which e.g. primary reserve from the unit can contribute to secondary reserve.

Set 'sft' does not contain 'ft_realized' which caused some important variable fixes not being made. Fixed by adding new sets 'sft_realized' and 'sft_realizedNoRseset' which include the realised t's.

Note: Smoothing for ts_node is not implemented yet

Use --input_dir=<path> and --output_dir=<path>. Defaults are ./input and ./output, respectively.

Activate this feature using active('scenRed') and setting mSettings parameters 'red_num_leaves' or 'red_percentage'.

Samples can now be used to describe uncertain future after the forecast horizon

Implementing a settings check that aborts execution if there are more efficiency levels defined in the model than there are in data.
Also adding abort messages to existing data and setting checks, as well as removing 'unit_current' as an unused asset.

Updating equation q_boundCyclic. The user can now bound the state variable to be the same in the first and last time intervals of any two samples. Defining "consecutive" and "parallel" samples in gnss_bound in the input data can still be difficult, especially since samples are model specific and defined in modelsInit.gms after reading input data. Making this more robust requires some further work.

Removing the "t_activeNoReset" by including "tmp_dt" historical time steps into "t_active" and avoiding loops over the entire model history.
Implementing a new "unitCounter" set to limit excessive looping over the "counter" set in unit startup/shutdown constraints.

Renaming "df_nReserves" to "df_reserves", changing "mft_nReserves" to "ft_reservesFixed", as well as a lot of accompanying changes.

Updating efficiency levels so that level1 does not need to be the first active efficiency level and it is possible to leave efficiency levels undefined between defined efficiency levels. If several efficiency levels have the same horizon, the last one is kept and others are removed. Note that unit aggregation should not start before t_jump.

Updating q_startshut equation to take into account unit aggregation. Work is still needed to take into account minimum online and minimum offline times of aggregator units. Unit aggregation most likely does not work properly with start-up and shutdown trajectories at the moment. The user should make sure that those trajectories are omitted before unit aggregation starts.

See issues #8 and #24.

Created two supersets 'uft_onlineLP_withPrevious'/ 'uft_onlineMIP_withPrevious' to the subsets 'uft_onlineLP'/ 'uft_onlineMIP' to include t00000 in the sets.

Reserve length is now a node property instead of a model property. In addition each reserve category can be turned on or off in the model settings.

Aggregation of units now handled by unitUnitEffLevel in inputData instead of 't_aggregate' in mSettings. Allows to choose aggregation threshold by unitAggregate and links it with the efficiency levels (and probably avoids issues since it was previously possible to have different horizons for aggregation and efficiency levels). Addresses issue #22.

Also bug fix to v_onlineMIP to v_onlineLP transfer where v_online was active for both at the t just before the switch.

Aggregation of units now handled by unitUnitEffLevel in inputData instead of 't_aggregate' in mSettings. Allows to choose aggregation threshold by unitAggregate and links it with the efficiency levels (and probably avoids issues since it was previously possible to have different horizons for aggregation and efficiency levels).

Also bug fix to v_onlineMIP to v_onlineLP transfer where v_online was active for both at the t just before the switch.

Output the name of the unit that is causing the model to abort when checking op parameters of units. Addresses issue #6.

* taking investment costs into account in cost results calculation
* adding transfer link investment possibility sets
* moving run-up and shutdown trajectories into their own 'blocks' to speed up the model
* commenting out some forecast related sets that are not used anymore
* updating info file writing

Implementing maxFuelCons and maxFuelFraction. For main fuels, the parameters function as their names suggest; for start-up fuels, maxFuelCons is not implemented and maxFuelFraction is treated as exact fraction.

Include t00000 for storage and online variables for the results output (as long as there are variable values for the first time period).

Adding ramping costs. Upward and downward ramps separated into different tiers (or 'slacks') with different costs.

Also changing the length of the forecast horizon so that it is now possible to either keep a static length for the forecasts or to have a decreasing forecast horizon that only extends forward at every tForecastJump.

Mainly updating the calculation of dtt and dt_toStartup. Also fixing two small but significant mistakes in the equations.