Introduced a new set `mst` with combinations of models, samples and t's.
Using projections in for calculating some lower-dimension subsets.

Previous solution was slow when there were many s's.

Poor performance when there were multiple samples used.

Also removed filter config files as they should be local.

Spill variable `v_spill` is actually treated as spill rate (MWh/h) in the equations, so fixed the description.
Result parameter `r_spill` is expressed in absolute energy, so rate must be multiplied with the step length.

- q_transferTwoWayLimit1 is used if investments not enabled
- q_transferTwoWayLimit2 is used if investments enabled and initial capacities are the same in both directions

If investments are enabled and initial capacities are not the same in both directions, a similar constraint would become nonlinear.

Issue #158

Issue #159

Adding a check to ensure that the start-up fuel node is in gnGroup.

Issue #156

r_gnnTransferMarginalValue updates: checking correct end node for marginal value with leftward and rightward transfers. Multiplying with trasferLosses. Removing unnecessary 10e-6 scaling.

Updates related to the loop over group reserve horizon when recording results....

Closes #148

See merge request !18

- Converting samples and time steps into indices when storing p_s_discountFactor, sGroup, gnss_bound, uss_bound, t_invest, and utAvailabilityLimits.
- Storing sGroup, gnss_bound, uss_bound, t_invest, and utAvailabilityLimits as boolean parameters of group, grid__node, unit, model, and unit object/relationship classes, respectively.
- Adding a 'model' object class for t_invest and p_s_discountFactor.

Does not help in my version of GAMS (24.1), since PUT writing facility fails with this. Hopefully works with newer versions.

Import
- not importing f objects anymore, f is now used as an index
- deleting all 'ts_...' relationship imports, including timeseries parameters in other relationships instead, using Base_ts alternative
- importing node boundaries as mappings

Transformer
- not renaming parameters anymore

Export
- updating the exporter according to the importer changes
- exploiting the new feature to export multiple mappings to the same GDX "table"
- filtering value types when exporting constants and timeseries/mappings

The alternatives should be used differently.

The diff links were not maintained.

Updating SpineDB to Backbone input GDX Exporter. Works with the current Backbone version, at least if there is data in every set and parameter.

Creating a new set groupRestype(group, restype) for better control of reserve-related assignments because commit cdf2b7c2 may have caused errors in certain special cases.

Issue #153

Updates related to the loop over group reserve horizon when recording results. The loop used to be very slow for large f, t sets.

Fixing issue #151: adding extra conditions to default value setting of both p_groupReserves(group, restype, 'update_frequency') and p_gnReserves(grid, node, restype, 'update_frequency').

Fixing issue #148: replacing '<' with '<=' in all 'ord(t) < tSolveFirst + ... restype, 'reserve_length')' in 2d_constraints.gms.

Add the information about superposed nodes in modelinit_temp.gms. Should probably be included in the input data at some point.