Rework Components query to query multiple dimensions at once

[sosiology]

Situation

• The Components GQL query queries multiple dimension individually as of now.
• The Elements with duplicate predicates (i.e. schema:identifier) figure twice in the result list.

Motivation

• The SPARQL endpoint is generally better performing upon recieving less queries - but optimized ones.

DoD

• [ ] The Components Query makes one SPARQL query over all dimensions at once (instead of multiple)
• [ ] The resultset has no duplicates regardless of duplicate predicates

[bprusinowski]

Hi @Rdataflow, I merged the #1487 a while ago to be able to compare the performance after the latest query improvements on the same infrastructure (INT data source on Visualize TEST and Visualize INT, with server-side cache disabled).

After a bit of testing, these are the rough results (times for the longest components query from the debug panel):

• BLW cubes: ~20s on TEST and ~5s on INT,
• Traffic noise pollution: ~1s on TEST and ~0.5s to INT,
• Stapfer: ~10s on TEST and ~25s on INT,
• Elcom: ~20s on TEST and ~40s on INT.

It looks that for some of the cubes the benefit is obvious (bigger cubes), but for smaller ones the regression is also clear. Let me know what you think about this 👀

cc @adintegra

[Rdataflow]

the long time it takes for BLW cubes is related to https://github.com/visualize-admin/visualization-tool/issues/1658