Utopya

- !68 lets the `xr.DataArray` underlying `XarrayDC` inherit attributes from the container instance (don't know why that wasn't configured to be the case before).
- !67 fixes a regression in the Read-The-Docs configuration.

- !66 fixes a bug in `ModelBenchmarkMixin` that prevented reconstructing a labelled DataArray for evaluation of the cumulative benchmarking results.

- !62 adds the `-W`/`--num-workers` option to `utopya run`, a shortcut to more easily set the number of worker processes.
- !63 fixes a YAML loading issue in the `MinimalModel` introduced by `ruamel.yaml >= 0.18`.
- !65 adds test environments for Python 3.12

- !61 fixes a bug in graph animation plots that prohibited plotting if additional graph attributes were set.
- Also reduces the verbosity of the graph plot by hiding repetitive log messages.
- !60 allows controlling how many tasks the `WorkerManager` spawns each iteration of the working loop, improving CPU utilisation when many short simulations need to be started in fast succession.

- !58 changes the default behaviour of the `utopya_backend` `BaseModel` to also seed numpy's `np.random` default RNG and the `random` module's default RNG when setting up a model.
This makes Python simulations deterministic even in cases where an external RNG is being used; *not* having done this in the past meant that simulations would not be reproducible, which is why this is considered a *bug fix*.
The behaviour can be deactivated by setting `seed_numpy_rng` and `seed_system_rng` parameters to False.
- !59 removes redundant YAML-related code, which is now implemented in the [`yayaml` package](https://gitlab.com/blsqr/yayaml).
- !59 requires [paramspace v2.6](https://gitlab.com/blsqr/paramspace), which includes improvements, bug fixes, and (minor) breaking changes.

- !55 improves graph layouting by using the pygraphviz package (instead of the outdated pydot).
For plotting networks, the utopya installation now includes a set of optional dependencies, installable via `pip install utopya[opt]`.