Output binning

[JanWeldert]

• Add global auxiliary data dict to container set, to make it possible to share keys between different containers. This can be used to share the output binning and regularized output binning.
• Check now if the output binning is the same as the apply mode of the hist stage(s).
• hist and snowstorm_hist stage can have apply_mode = None. In that case they will directly take the output binning.

Using the regularized output binning from the hist stage in the snowstorm hist stage speeds up the latter's computation.

It is important to enforce the apply modes of the different stages (especially the hist stage) to be the same as the output binning because re-binning is not currently supported by PISA. In principle, having a different hist stage apply mode and output binning would be possible and produce the correct results. However, it would not make sense to bin the events in the hist stage just to re-bin them immediately in order to obtain the output map.

[thehrh]

Just a handful of small remarks to address.

It will be important to adapt and improve (where necessary) the two notebook tutorials on stage modes and eventwise information in an upcoming PR.

[thehrh]

I just fixed the one outstanding review comment. Looks good to me now.