feat: save/fetch Reporter as DataSource

[victorgarcia98]

This PR introduces the ideas discussed in Discussion #662:

• Store/Fetch Reporters from DataSources (config + class)
• Rename reporter_config to config. This will keep static parameters of a reporter, the ones that we want to persist to the DB.
• Introduce inputs to keep parameters that change, e.g., start, end, etc. In the discussion we refer to this set of parameters as the report_config.

Apart from that, this PR adapts the previous work to the DataGenerator class

Creation of the DataGenerator class

• Base class for Scheduler, Reporter and Forecaster
• Static configuration is defined in _config and (de)serialized with _config_schema.
• Supports passing configuration through the keyword arguments in the DataGenerator constructor.
• Dynamic parameters are defined in _input and (de)serialized with _inputs_schema.
• Supports passing inputs through keyword arguments when calling the method compute. For example, reporter.compute(start=NL_tz.localize(datetime(2023, 1, 1)) is equivalent to reporter.compute(inputs=dict(start="2023-01-01T00:00:00+02:00"))
• Store the available DataGenerators in the app context, in the `
• Includes the _config into the DataSource attributes.
• Fetch a stored DataGenerator + config from its corresponding DataSource using the data_generator property.
• Reporters, Schedulers and Forecasters registered in a single application context property, data_generators. The class is defined in the model field of the `DataSource.

Adaptation of the Reporters

This PR also adapts the reporters introduced in v0.14.0 to work as DataGenerators.

• Reporter
• PandasReporter
• AggregatorReporter
• TibberReporter

[Flix6x]

I did a line-by-line review and got through most of it. Only test_reporting.py and test_data_source.py left. Next part of my review would include some testing (e.g. how to adapt previous reporter config JSONs to keep functioning, assuming an adaptation is needed at all).

Let's also have a quick discussion this week on the idea of allowing sensor IDs to be set in the compute input rather than in the __init__ config.

[Flix6x]

Let's triage my comments together to see what should be part of the PR still.

[Flix6x]

I see no new commits except a merge from main?

[Flix6x]

Nice, thanks!

(For anyone reading along, we'll tackle the schema field renaming in a new PR.)

[Flix6x]

(And the changelog entry is being added in the PR #752.)