The german website Wikipedia is referencing, is mostly about SSTCs.
A useful figure of merit is the power transferred to the secondary
for a given input current. The input current is a mixture of a resistive
and a reactive part, the latter usually being much larger. The resistive
part is the one transferring power into the coil and that part peaks
at the secondary resonance frequency. By choosing a frequency
somewhat beside the resonance, it is possible to reduce the reactive
part of the current, but the optimum amount of detuning depends not
only on
the coupling, but also on the loading of the secondary.
A much better way to reduce the reactive part is by putting a cap in
series
on the input side, such as in a DRSSTC. In this case the preferred input
frequency is the secondary resonance frequency and the necessary cap
for this must be chosen with L1*C1=L2*C2. Sadly this configuration
has stability issues. Stable operation can be obtained at the pole
frequencies.
There L1*C1=L2*C2 might not be the best choice, but that is a topic
way outside the Wiki article.
For an SGTC, L1*C1=L2*C2 is necessary for a complete transfer of
primary energy to the secondary.
The Wikipedia entry defintely needs a correction.