Best practice for dynamic connection selection in Fabric Data Pipelines

I think in general you thinking is sound, I would only store the connection name in your config and use the Fabric API to get the connectionid as part of your pipeline, this will make the solution more portable as you move through environments without having to note and store all of the connection ids.

As I am sure you know you will have to create seperate copy activities for each source type within your pipeline with a switch activity to route the operations or seperate pipeline for each type of data source you are copying data from.