Fixing issue #32. The last time interval in the run-up phase had a wrong limit for the q_rampUpLimit, this was corrected by calculating a new parameter p_u_maxRampSpeedInLastRunUpInterval. This might not be the only thing wrong. Needs to be checked futher.

A fix to issue #51: uft_online_withPrevious now includes the interval just before the start of a sample --> online status is free in the first interval of each sample

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

Changed 'q_boundStateMaxDiff' to correctly measure difference in state variables, and not the state energy content.
Reserve provisions shouldn't be included if 'energyStoredPerUnitOfState'=0.

Issue #50: Changed the 'mStart' values in the model definition file templates to avoid errors with the simulation start time.

Default penalty changed to 1e9.

Changing the 'p_nuRes2Res(node, unit, restype, up_down, restype)' description and function so that the first 'restype' also provides the second 'restype'.
Also moving the 'r_reserve2Reserve' calculation from "4a_outputVariant.gms" to "4b_outputInvariant" for efficiency.

Output reserve2reserve (included in r_reserve, but there is also separate r_reserve2Reserve that should show just reserve to reserve provision). Should be tested.

Aggregated groups appeared in the results although the model was not using the effLevel where the aggregation starts. This was fixed by moving the set creation for aggregated units to 3a_periodic_Init.gms (because it needed mSettings that has m indice) and changing the way unit_aggregator, unit_aggregated and unit_noAggregate sets are defined (units that are aggregates will not appear any more at all if they are not defacto in use).

Included the 'N-1' parameter into 'param_policy' to avoid errors if no "p_nReserves(node, restype, 'N-1')" are defined.
Additionally, reorganized and highlighted the reserve related parameters in 'param_policy', as they hopefully will be moved to a dedicated parameter set at some point.

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.

Fixing an issue, where the averaging of time series data would not work properly for 'stepLengthinInHours' != 1.

Minor readability and performance upgrades on fixing the previously realized startup/shutdown/online variables.

Further fixes to 'uft_onlineLP_withPrevious' and 'uft_onlineMIP_withPrevious' and fixing the historical startups to allow for arbitraty 'stepsPerInterval' on the first interval.

Minor fixes to 'tmp_dt', 't_active', 'ft_realized', 'ft_realized', 'ft_realizedNoReset' and 'msft_realizedNoReset' to allow for arbitrary 'stepsPerInterval' immediately on the first interval.

Adding 'reserve_length' to 'param_policy' to avoid missing label errors with input data without reserves.
Also a minor clarification to 'q_conversionDirectInputOutput'.

Removing 'dt_noReset' as unnecessary, now that 't_active' and 'dt' cover the required model history.
Also fixing the calculation of 'dt' for the first time step in 't_active'.

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