Merging Digest::SHA3 into Digest::SHA is inadvisable for the following two reasons:
1. Current NIST policy encourages the use of SHA-2 (i.e. SHA-224 up to SHA-512) for applications employing secure hash algorithms. All of these algorithms are implemented in Digest::SHA. NIST also remarks that SHA-3 *may* be used, but "at this time, there is no need or plan to transition applications from SHA-2 to SHA-3." In other words, use of SHA-3 is optional at this point. This is certainly a wise policy, given that current field experience with SHA-3 is minimal, and that SHA-2 is still more than adequate in providing very high levels of security.
2. Modularity is always more desirable, especially when various components are optional and users might wish to ignore them. Digest::SHA is already a very large and complex module, which it needs to be to implement the full NIST SHA-2 standard. Burdening it further with extraneous, optional capabilities is ill-advised, particularly given that it's now a part of the Perl core.
But I appreciate your suggestion. Interest in SHA-3 was extremely keen when NIST first announced its selection of Keccak. So I worked very hard to produce a full implementation of the SHA-3 standard, including BIT capability and the so-called SHA-3-0 variant which allows for variable-sized hashes. Once the novelty of the announcement wore off, interest in SHA-3 dropped to virtually zero. And that's not surprising, given that SHA-2 is faster and still highly secure.