FoReco 1.2.0

• (CRAN Package Check Results) Fixed compatibility issues with osqp 1.0. The package is now fully compatible with any version.
• Added csmvn(), temvn(), and ctmvn() for Gaussian probabilistic forecast reconciliation in the cross-sectional, temporal, and cross-temporal frameworks using the distributional package;
• Added cssmp(), tesmp(), and ctsmp() for sample-based probabilistic forecast reconciliation in the cross-sectional, temporal, and cross-temporal frameworks using the distributional package;
• Added as_ctmatrix() and as_horizon_stacked_ctmatrix() functions to convert between horizon-stacked (cross-temporal version) and cross-temporal layouts;
• Added as_tevector() and as_horizon_stacked_tematrix() functions to convert between horizon-stacked (temporal version) and temporal layouts;
• Added non-negative forecast reconciliation algorithms bpv (block principal pivoting algorithm), nfca (negative forecasts correction algorithm), nnic (iterative non-negative reconciliation with immutable constraints), and sntz (set-negative-to-zero with bottom-up and top-down alternatives) based on:
  • Girolimetto, D. (2025), Non-negative forecast reconciliation: Optimal methods and operational solutions. arXiv;
  • Kourentzes, N. and Athanasopoulos, G. (2021) Elucidate structure in intermittent demand series. European Journal of Operational Research, 288, 141-152. doi:10.1016/j.ejor.2020.05.046;
  • Wickramasuriya, S. L., Turlach, B. A., and Hyndman, R. J. (2020), "Optimal non-negative forecast reconciliation", Statistics and Computing, 30(5), 1167–1182. doi:10.1007/s11222-020-09930-0;
• Added ... for simulate() additional arguments in csboot(), teboot() and ctboot();
• Fixed bugs and improved stability.