Reimplement rd_sum.pandas_frame() to use resfo

[berland]

Following #1062 a natural next step is to similarly reimplement

https://github.com/equinor/resdata/blob/main/python/resdata/summary/rd_sum.py#L563-L624

using resfo. New code needed compared to #1062 is time interpolation.

[eivindjahren]

Because rd_sum reads the files on __init__ I think it would be not so useful to read the entire summary again with resfo. However, we could have code that does the equivalent of lambda case_path, time_index, column_keys: Summary(case_path).pandas_frame(time_index, column_keys). Would that work for your purposes @berland ?