"Please describe how 'time-compressing' a waveform is different than frequency-shifting it"
If I frequency shift a waveform by a factor of 2, then the time compression is also a factor of 2. The article doesn't really mention it, but the frequency shifts in this experiment are much less than a factor of 2, but the time compression is from 2.5 ns to 95 ps, a factor of 27 compression.
This is a real time lens. A spatial lens works by imparting a quadratic spatial phase to light. Diffraction then causes the beam to focus due to the quadratic spatial phase.
A time lens works in analogy to a spatial lens by imparting a quadratic temporal phase to a light pulse. Propagation in a dispersive media then leads to the time compression.
The difficulty is it is very hard to impart a quadratic phase to short light pulses. The only real way to do it is nonlinear optics, which is where the (small) frequency shifts mentioned in the article come from.