Fix Aggregator setter

[schmoelder]

Fixes #215

Added methods to indicate whether the internal data should be transposed (i.e., have shape (n_containers, *parameter_shape)). While this resolves the immediate issue, I had to overwrite the __set__ method of the Aggregator base class, which feels less than ideal. I'll revisit this implementation to explore alternative approaches, but it should work for now.

Additionally, I noticed there are currently no tests for any aggregated parameters. This is unexpected, as I was fairly certain I had written some. Maybe they got lost somewhere...

ToDos:

• Stoich coeffs in reactions should be sized
• Check if the wrong code snippet is commented out in SizedAggregator.__set__
• Check length of container vs value in Aggregator.__set__
• Check consistency of transpose=True vs False in SizedAggregator
• Add tests

[schmoelder]

I refactored the transport functionality and now only the __get__ and __set__ methods are extended in the subclass (instead of being overwritten) which I find much more elegant.

I also added tests for the aggregator, so once tests pass, I'd appreciate another thorough review. 🤓

[ronald-jaepel]

Usage in optimizationProblems works nicely. 👍

Setting slices of aggregators didn't work. It didn't raise an Error or Warning and also didn't change the values. I've extended the tests for that. If that is a use-case you don't see users using, feel free to remove the tests again.

[schmoelder]

Setting slices of aggregators didn't work. It didn't raise an Error or Warning and also didn't change the values. I've extended the tests for that. If that is a use-case you don't see users using, feel free to remove the tests again.

Damn, that's a tricky one, since we're always returning a new value that is being constructed by iterating over the internal instances. So there is no internal reference that points back to the underlying container elements. I'm honestly not completely sure how to handle this but I'll think a bit further about it.

[schmoelder]

Puh, this was a "difficult birth". In the end I did decide to provide proxy objects to the aggregated parameters which now seems to work properly. I had to adapt one more test but I think it's fine now...

[schmoelder]

This one here caused a lot of trouble... 😅