Hi everybody, I'm a bit late to the party but I think you're describing my number one frustration with dashboards.
One of my core datasets is a property database with multiple layers representing land, access rights, parking, communal areas, etc. each linked using a property ID. I would like the user of the dashboard to select the property name, which is only stored in the land layer, then filter the other layers using the property ID. The dilemma I face is as follows:
The land layer can contain multiple non-merged polygons, so if I use the Categories from Features option, I get duplicate values displayed in the selector, but at least I can specify the join field, and the filters work.
If I use the Grouped Values option to get around the duplicate values issue, I have to use the property ID as the display field in the selector in order to join to the other layers. I then either have to get the user to memorise what every property ID relates to or manually override 1000s of labels so I can use the property name (also why can't this be chosen from another field?).
This is just one example of how this problem manifests itself and I have many other projects where the same issue arises, and I often resort to option one above as the frustrating solution. Admittedly I haven't got as far as some of you looking into creating related tables and expressions (I don't have the time), but this feels like it ought to just work out of the box.