Introduce support for querying Archetypes

[jleibs]

What

The main change here is to introduce a new ArchetypeView that uses the archetype interfaces as an alternative to the historical EntityView.

Because EntityViews and legacy type definitions leak through numerous APIs, this involved forking a number of APIs:

• process_archetype_views
• query_archetype_with_history
• query_archetype
• range_archetype
• process_color_arch
• process_radii_arch
• process_annotations_and_keypoints_arch

As we migrate more of our queries to the new APIs, the legacy versions of these helpers can be incrementally deleted and potentially renamed as appropriate.

Future work

This does not yet replicated the query_tests and range_tests for Archetype APIs. Part of the reason for this is that we make heavy use of DataCell construction in those tests, but DataCells are still heavily dependent on the legacy Component APIs.

Checklist

• I have read and agree to Contributor Guide and the Code of Conduct
• I've included a screenshot or gif (if applicable)
• I have tested demo.rerun.io (if applicable)

• PR Build Summary
• Docs preview
• Examples preview

[teh-cmc]

Man, this unburied a lot of repressed memories :D

So many opportunities for clean up, optimizations, smarter traits etc in the future 🤯

[teh-cmc]

Interesting; I think it's becoming quite clear that we want the Archetype trait to be split into two: the static part vs. the data/arrow part.

E.g.

pub trait ArchetypeInfo {
    fn name() -> ArchetypeName;
    fn required_components() -> Vec<ComponentName>;
    fn recommended_components() -> Vec<ComponentName>;
    fn optional_components() -> Vec<ComponentName>;
    fn all_components() -> Vec<ComponentName>;
}

pub trait Archetype {
    type Info: ArchetypeInfo;

    fn to_arrow(&self) -> Vec<(::arrow2::datatypes::Field, Box<dyn ::arrow2::array::Array>)> {
        self.try_to_arrow().detailed_unwrap()
    }

    fn try_to_arrow(
        &self,
    ) -> SerializationResult<Vec<(::arrow2::datatypes::Field, Box<dyn ::arrow2::array::Array>)>>;
    fn from_arrow(
        data: impl IntoIterator<Item = (::arrow2::datatypes::Field, Box<dyn ::arrow2::array::Array>)>,
    ) -> Self
    where
        Self: Sized,
    {
        Self::try_from_arrow(data).detailed_unwrap()
    }

    fn try_from_arrow(
        data: impl IntoIterator<Item = (::arrow2::datatypes::Field, Box<dyn ::arrow2::array::Array>)>,
    ) -> DeserializationResult<Self>
    where
        Self: Sized;
}

[teh-cmc]

Talk recap: let's make ArchetypeInfo an actual struct instead (using e.g. static slices or padded static arrays)

[teh-cmc]

Interesting; we could generate this Primary type as an associated type of the ArchetypeInfo trait.

[teh-cmc]

Talk recap: don't need to, we'd like to move away from single primary components (i.e. one BIG component with all mandatory stuff) and towards multiple required components (which are intersected instance-key wise)

[teh-cmc]

Those don't actually use the df_builder_arch things defined above 🤔 Anyway we can improve all of that later.