Atomica

- Improve numerical robustness of `SpendingPackageAdjustment` under certain edge cases
- Fix bug in cumulative plot labelling that could result in the axis label containing more than one 'Cumulative' prefix

- Allow initializing compartments and characteristics with a 0 value by setting a default value without needing to add the quantity to the databook. This simplifies initialization of models that have large numbers of compartments that should always be initialized with a 0 value, without needing to add many databook entries or extra initialization characteristics.

- Allow framework variables with single characters (previously, all code names had to be at least two characters long)
- Improve handling of automatic number of workers if a number is provided instead of a collection of inputs
- Add `optim_args` argument to `at.optimize` which allows arguments to be passed to optimization functions such as ASD

- The "Databook pages" sheet in the framework is now optional, if a compartment, characteristic or parameter has a "Databook page" that does not appear in the "Databook pages" sheet (or if the "Databook pages" sheet is missing entirely) then the specified page will be created with the specified name as both the code name and full name. As the "Databook pages" sheet is created and populated with these names during framework validation, downstream code expecting the sheet to exist should not require any changes.

- Fix array size error for junctions belonging to a duration group (some otherwise valid frameworks previously raised an error when running the model)
- Fix missing cells/NaNs in equivalent spending caused by numerical precision errors

- Unpin `matplotlib` version in `setup.py`