yes, GRD is already multi-looked. It is not a mandatory step, because before you can use SAR data in a GIS you have to geocode them which is done in SNAP during the Range Doppler Terrain Correction which also applies a resampling, resulting in square pixels anyway.
Multi-looking is good to make SLC data easier to handle but it makes little sense for GRD data unless you want to systematically alter the spatial resolution at consistent intervals.