Dashboard with chained, static variables

I am trying to set up a dashboard with panels that have 3 parameters - each a variable that can be deduced from 1 user selection.

Concrete: The user selects one “alert window duration” from a dropdown, which gives: {longwindow}, {shortwindow}, and ${burnrate}.

Here are the values that belongs together:

{ longwindow: 1h, shortwindow: 5m, burnrate: 14.1},
{ longwindow: 6h, shortwindow: 30m, burnrate: 6},
{ longwindow: 1d, shortwindow: 2h, burnrate: 3},
{ longwindow: 3d, shortwindow: 6h, burnrate: 1},

I had hoped to be able to implement this with chained variables and some kind of static dataset, but I had to reach out for a SQL-compatible datasource.

Currently, its implemented with three variables:
1: longwindow, type Interval, values: 1h, 6h, 1d, 3d
2: shortwindow (hidden), type query, datasource: SQL, query:

    when '$longwindow' = '1h' then '5m'
    when '$longwindow' = '6h' then '30m'
    when '$longwindow' = '1d' then '2h'
    when '$longwindow' = '3d' then '6h'
    else 'invalid'

3: burnrate (hidden): same as shortwindow, but returning the related burnrate values related to the given longwindow.

The variables is used in a set of Prometheus-queries: “metric_name[{shortwindow}]", "metric_name[{longwindow}]”, and “vector(100-0.01*${burnrate})”.

It works, but it feels kind of silly calling out to an external service to execute this.

I’ve looked through the available datasource plugins (and even tried one for static data) to find a simpler solution, but with no luck.

Can anyone see/suggest a solution that does not involve staging up and running this simple logic on a remote data source?