Update SIENNA link (#1611)

Author: datejada

paper/paper.md

@@ -93,7 +93,7 @@ TulipaEnergyModel.jl focuses on model quality and efficient implementation, allo
 
 ## Statement of Need
 
-Existing models and frameworks in Energy System Optimisation include [EnergyModelsX](https://github.com/EnergyModelsX) [@EnergyModelsX], [PowerModels](https://github.com/lanl-ansi/PowerModels.jl) [@PowerModels], [SpineOpt](https://www.tools-for-energy-system-modelling.org/) [@SpineOpt], [Sienna](https://www.nrel.gov/analysis/sienna) [@Sienna], [GenX](https://github.com/GenXProject/GenX) [@GenX], [PyPSA](https://pypsa.org) [@PyPSA], and [Calliope](https://github.com/calliope-project/calliope) [@Calliope].
+Existing models and frameworks in Energy System Optimisation include [EnergyModelsX](https://github.com/EnergyModelsX) [@EnergyModelsX], [PowerModels](https://github.com/lanl-ansi/PowerModels.jl) [@PowerModels], [SpineOpt](https://www.tools-for-energy-system-modelling.org/) [@SpineOpt], [Sienna](https://www.nlr.gov/analysis/sienna) [@Sienna], [GenX](https://github.com/GenXProject/GenX) [@GenX], [PyPSA](https://pypsa.org) [@PyPSA], and [Calliope](https://github.com/calliope-project/calliope) [@Calliope].
 However, they run into computational limits when solving large-scale problems and must resort to (over)simplifying the model to reduce computational burden. The common misconception is that the only strategy to speed up solving times without sacrificing model fidelity is through faster solvers or computers.
 However, the strategy that is widely overlooked is improving the quality of the mathematical formulations, which increases model fidelity while simultaneously solving faster than standard formulations.
 This insight inspired the development of TulipaEnergyModel.jl, with the core philosophy of advancing the state-of-the-art in formulation quality by: 1) lowering computational cost while maintaining model fidelity, by reducing the problem size [@Tejada2025], and by creating tighter mixed-integer programs (MIP) [@MoralesEspana2013]. 2) increasing model fidelity without extra computational cost, e.g., by developing more accurate linear programming (LP) approximations [@Elgersma2025; @gentile2016; @MoralesEspana2022]. Finally, 3) balancing computational burden with adaptive/flexible model fidelity, i.e., having different levels of detail in various parts of the model, in the temporal [@Gao2025], technological [@MoralesEspana2022] and spatial dimensions.