So in v1, if you selected to log types of "cname", and a request for an "a" came in, you would not see the request logged, but if a cname was part of the resolution, you'd see the one "cname" response record. It does lack any context of why its logged though.
In v2, setting the type to cname will only log DNS transactions when the request is for a cname. This was intentional, from the commit log, and I believe it was discussed in Jabber or something, but I didn't relay that discussion into the Github pull request.
With v2 logging we could do similar, its a little more complex. If you just selected "cname", you wouldn't get the request for the "a" that created the "cname" response, and the response would only contain the "cname" portion, and not the full response. OR, it could mean only log records with a "cname" entry, then it would get the full response, but still no discrete request object, which is fine, as the full response does have the query in it.
I think in v1 logging this feature primarily existed to cut back on the number of records. v2 logging cuts back on the number of records logged by design, and changes things enough that this feature is not straight forward to implement without some discussion on how it should be done.