Bug: Score of multiple models fails if observations are not in each comparer

[ecomodeller]

import modelskill as ms
import pandas as pd

time = pd.date_range("2000", periods=2)
cmp1 = ms.match(
    obs=ms.PointObservation(pd.Series([0, 0], index=time), name="foo"),
    mod=ms.PointModelResult(pd.Series([0, 0], index=time), name="bar"),
)
cmp2 = ms.match(
    obs=ms.PointObservation(pd.Series([0, 0], index=time), name="baz"),
    mod=ms.PointModelResult(pd.Series([0, 0], index=time), name="qux"),
)
# TODO should it be allowed to combine two comparers with different models, maybe not?
cc = ms.ComparerCollection([cmp1, cmp2])
cc.skill(metrics="rmse")

So far so good, but if I call

cc.score() 😟

it fails with a cryptical message since the weights vector is of the wrong length, if each comparers doesn't contain each observation.

        if len(data) != len(index):
>           raise ValueError(
                "Length of values "
                f"({len(data)}) "
                "does not match length of index "
                f"({len(index)})"
            )
E           ValueError: Length of values (4) does not match length of index (2)

This PR fixes this problem, and does some more refactoring related to weighted scoring.

[ryan-kipawa]

Looks good :)

I assume "should it be allowed to combine two comparers with different models, maybe not?" is just a general comment outside the scope of the PR.

[ecomodeller]

Looks good :)

I assume "should it be allowed to combine two comparers with different models, maybe not?" is just a general comment outside the scope of the PR.

Indeed, thinking out "loud"...