Re-implement more metrics

[bblodfon]

• Scoring rules: Practically we need to re-implement in C++ the Brier Score (IBS and IBLL) from Schoop 2011 (version from riskRegression, also I think the same version used in Kretowska 2018) and the extended version from Alberge 2025. Add more parameters in the function, the version from riskRegression is a bit limited. Also an RCLL for competing risks as mentioned in the MLSA book would be nice to have (needs transformation + interpolation from CIF(t) => $f(t)$ and $S(t)$.
• Time-dependent C-index => Antolini's C-index for competing risks as in DeepHit (2018) paper, see code (I think Kretowska used the same, Bender 2021 used this as well, wrapping it in R). This paper uses the Cum. hazard instead. Probably a variation of Gandy's C-index for crossing hazards when the translation from CIF(t) => $h(t)$ hazards is available.
• Blanche's AUC(t) from the 2014 paper (version from riskRegression) - re-implement, "broadening" up a bit the interface.
• Calibration metrics, see Alberge 2026 - I see CR D-calibration and $cal_K^{\alpha}$ there.