LOL. I remember lusting after a GR 1603A.. until I got one. Odd, indeed!
It is difficult to specify really low impedance measurements as % because "small" errors (from varying contact resistance of fixture and connectors, the impact of thermal changes on the instrument, DUT, fixture, etc.) become an increasingly large % of the measurement. Look at the specs for any LCR/Z meter and you'll see the accuracy specs loosen dramatically below say 100mOhm. Because of this, for low impedance measurements, I tend to look for repeatability first, and then accuracy as a "repeatably resolvable increment".
For the sake of discussion, let's say you want to measure down to a 1mOhm lower limit, with 100uOhm resolution, from (some low freq) up to 100kHz. (I choose 100uOhm resolution because my experience is, that is about the lowest repeatable resolution you can get with a high-quality bench DMM using 4-wire Ohms, and most won't even do that repeatably.)
I also sense that you want more freqs than just the spot decade ones (100, 1k, 10k, 100k).
So, of the multi-frequency oldies I've used that can be found for reasonable money :
HP 4276A meets the criteria, but only up to 20kHz.
Wayne Kerr 6425 comfortably betters it on all counts.
Ditto the ESI 2150/2160, but they're clumsy for bench use (user interface on the front, measurement terminals on the rear).
GenRad 1689x and 1693 models, but they are the darling of industry and still command higher $$.
With some math and programming, you can also do this with any decent DSA or VNA. Bring the measurement data into a computer and do the error correction and impedance conversion there.
You'll need either floating or isolated inputs, or isolated output (xfmrs work) so ground loops don't impact the measurement.
For 100kHz, the HP 3562A would be an excellent choice. You can float the inputs, and it has enough working registers and programming capability to be able to do the calculations on the fly and plot the results in "user units". (I use the 3562A's PC-based sister 3567A for this with excellent results.)
So, measuring low impedances up to 100kHz requires some care but is very doable.