Unlike for regular read-only views, it makes sense for time-series collections to support the collStats command. Most of the stats would come from the actual bucket collection, except for count (unfortunately it's not currently possible to store that for a time-series collection), the derived avgObjSize and storage engine stats, such as the very verbose wiredTiger section: for that level of detail it makes sense to directly get stats from the buckets collection.
Additional stats could be bucketCount (which is count of the buckets collection), and avgBucketSize and possibly bucketNs or similar. A question is whether to include execution stats, such as the number of insert operations vs updates, measurements per commit, buckets closed due to size/count/time constraints, new buckets created due to time going backward, new metadata etc. Some of the more global bucket catalog information may be better included in a new serverStatus section instead.