In general a Bucket is synonymous with a Database in an RDBMS, so you would typically put multiple Tables -worth of data in a single Bucket.
Considering your workload, in addition to the number of documents in the system you also want to look at mutation rates - that is generally the constraining factor in Views performance, as each Design Document needs to consider (i.e. run the map & reduce functions on) every change which occurs in a bucket.