Adjusting base_score for BaseBoostingAssembler and XGBoostTreeModelAssembler

[NinaHerrmann]

First of all thank you for the library, I really appreciate using it. I am not to familiar with python libraries so excuse me if I am missing something obvious.
Problem:
When generating XGBoost models I came across the error:

File “[…].venv/lib/python3.13/site-packages/m2cgen/assemblers/boosting.py", line 80, in _assemble_bin_class_output
    base_score = -math.log(1.0 / self._base_score - 1.0)
                           ~~~~^~~~~~~~~~~~~~~~~~
TypeError: unsupported operand type(s) for /: 'float' and 'NoneType'

XGBoost 2.0 New says:

In the previous version, base_score was a constant that could be set as a training parameter. In the new version, XGBoost can automatically estimate this parameter based on input labels for optimal accuracy. (#8539, #8498, #8272, #8793, #8607)

github

Proposed solution:

To be compatible I check for None in BaseBoostingAssembler and check the two different ways (model.get_params().get("base_score") and model.intercept_ is not None) in XGBoostTreeModelAssembler passing it to the super.

Related:
#587 #581

Thanks for your feedback!