Custom question: Allow Filter before Join

Nobody has claimed this yet.

## Problem
I have a questions in which I join a quite large dataset (~100M records) with a lookup table. However, I am only interested in a subset of the base database (say 1k records).

The current query takes quite some time (> 1 minute), and while I am also improving other parameters (db hardware, cache, indexes, materialized views, etc.) I want to streamline the join operation.
For that, **I would like to filter records from 100M to 1k, *and only then* join on 1k** — rather than join on 100M and then filter to 1k.

## Expected solution
**In the current GUI (v0.34.1), you can only filter the result of all the join operations** (i.e. join on 100M, then filter to 1k).

What I would like is to be able to:
1. Import a base data set (100M)
2. Apply a filter on this data set (100M ￫ 1k)
3. Join the filtered base data set with another (e.g. lookup) table (1k)
4. [… continue with the rest of my question]

## Workarounds
Well, I can either:
* Filter after all the joins have been done (current natural solution): but this creates a resource heavy query, what I want to avoid;
* Import the base data set, apply the filter, create an aggregate that will actually return all the records, then join with the lookup table: this is just bulky, clunky, and feels wrong;
* Use a SQL query: but my point is to streamline the GUI query, without having user rely on SQL.

## Importance
I believe it's rather be a nice to have, as it's not blocking my workflow — just make things slow and my db suffer.

:arrow_down: Please click the :+1: reaction instead of leaving a `+1` or `update?` comment

Start by tracing the GUI query flow for filters and joins, then determine where filter placement is represented before query generation. Done means users can apply a base-table filter before joining, the generated query avoids joining excluded rows, and the existing join workflow still works.

Written by the indexing model from the issue text.