Why not use Norm[...,p] on the SingularValueList?

Ad sign of the x coefficient: I don't know of any trick simpler than multiply the result by –1 if you don't like it.

This is not even related to Sum. Compare: Integrate[2^(-1 + s) 3^-s + 2^(-1 + s) 5^-s, s] — Integrate[2^(-1 + s) 3^-s, s] — Integrate[2^(-1 + s) 5^-s, s] (the first integral is simply wrong).

Given that I suppose you are going to install the new version when you can use it, why not install it anyway and just try? If the license is not ready it will tell you and refuse to start, but you can still run the previous version as long as you used two different installation directories.

Thank you, @ilian. I assume it's the same case for option AllowHeads of Dimensions. Should I report when I stumble upon similar cases?

Nice answer! I just think in calculating periods you are reinventing Differences.