That is, the detecting is done by eddy currents in the target changing
the resonant frequency of the detecting coil.
The question doesn't become what you think it does because many metals will have resistive losses easily detectable with hardly any discernible change in the frequency of oscillation of a tuned search coil.
The resonant frequency will also change (with virtually zero eddy currents) when a piece of ferrite is brought into the search head field.
At hundreds of kHz, stainless steel produces both ferromagnetic and eddy current disturbances in the search head field and, these two effects drive the oscillation frequency in opposite directions hence, there can be zero net change in the frequency yet quite sizable resistive losses detectable.
However, if you compared a flat sheet with a multiturn shorted coil the coil will usually win because the flat sheet has the inductance of 1 shorted turn yet (say) 5 turns in the coil will have a leakage inductance 25 times bigger. In other words it will have a bigger disturbing effect on the field.