Ability to add sensor capacity if they measure power

[nhoening]

This is useful in general, but especially for getting started and then seeing your power sensors on the dashboard.

[Flix6x]

Honestly, I don't particularly like adding this custom logic to support a single attribute, also because it's an attribute that we intend to replace by a sensor of itself. I have an alternative suggestion:

@click.option(
    "--attributes",
    required=False,
    type=str,  # a json string, actually (maybe there is out-of-the-box validation for that)
    help="Additional attributes. Hint: for sensors that measures power, use {"capacity_in_mw": 10} to set a capacity of 10 MW",
    default = "{}",
)

[nhoening]

We have logic relying on that attribute being there, so it's not weird to also have logic to make sure the power sensors have it when being made.
Even if we support all potential attributes to be passed in with JSON (for which I haven't found any click support), I'd still want that check.
That code which relies on the attribute and this new code are both to-be deprecated, I think that's fine.

[Flix6x]

Okay, I like this better, and just as a placeholder. I'd like us to follow up with a ticket to replace any logic using the capacity_in_mw attribute, such that it becomes a completely optional attribute. I believe it's at least used to set the scale range on the latest state plot, and by schedulers to set constraints on the power flow. For the former we can use a max on the history, or let it go altogether (let the chart determine the scale range), and for the latter we can let constraints be set through a dedicated UDI event field (an optional field, in case the sensor attribute is set as a fallback).

[nhoening]

I agree. I'll make the ticket.

As a slightly longer outlook: For the former (and the latter maybe as well) we can also first look for a dedicated capacity sensor, then fall back to something more simple and robust.

In 2022, I think we need to make sure the new data model is super straightforward to use, as well as robust.