Server-side-proposed partial clone configuration (default clone --filter option)

 From a (GUI) client perspective, it would be quite helpful to propose 
good (partial) clone defaults to the user: whether to do a full clone, 
or skip large blobs, ... this can result in much smaller clones by 
*default* and thus a better user experience.

Such a proposal should most likely be a property of the repository 
itself, for example stored in a specific ref which can efficiently be 
fetched using Git, in preparation of the main clone.

Are there any best practices on how to implement that? Or even better, 
is there any specification on where this meta information should be 
stored, how it should look like, ...? If not, is anyone interested in 
this topic? I feel that this is something which not every 
platform/client should reinvent.

-Marc