Data for England has been supplied by Natural England under the Open Government Licence with the express intention of it being used by OSM. Data for Wales is from rowmaps.com with licensing statements as per that site and Robert Whittaker’s documentation.
I would suggest that we shouldn’t be tracing these as such*, but instead trying to map what is physically present first, and then using the PRoW data to tag physical paths that align with Rights of Way with appropriate designation=*, prow_ref=* and access tags.
In some of cases, you might be able to use a Rights of Way overlay together with aerial imagery to infer a physical path, but there’s still a risk that you might miss an impassible ditch or barbed wire fence at some point. You also miss the opportunity to tag barriers (e.g. stiles and gates) and steps that might be important for some users of the routes.
* unless of course you want to capture the route of the public right of way, independently of what paths there are on the ground. But it’s probably better not to do this unless the actual paths already been mapped in OSM and clearly deviate from the Definitive Line.
Yes, I was eliding how to use them, but generally I’d agree they’re not suitable for unthinking tracing. The two scenarios I find them most useful for are (a) adding designation/access to ways that have been mapped with unspecific tags (typically highway=service, track or path) and (b) small connectivity fixes.