odach_cc stratification reproducibility issues

[afarnudi]

The following issues will be addressed for the ODACH_CC pipeline:
1- Application of Prentice weights for left truncated survival data
2- removal of sites where the survival::coxph() breaks down during the initialisation step due to lack of events
3- Addition of stratification in the survival::coxph() input formula
4- Definition of new gradient and hessian matrixes extracted from the survival::coxph_fit object
5- Fix dimensional discrepancy between the aggregated D2 matrix and the derived one in the estimation step.

[afarnudi]

I will resolve these issues in my Fork and put up a pull request.