I am not familiar with this situation, so consider the possibility that my lips are just flapping in the wind. I have played with large magnets in the past, and currently work with electric motors at the design stage.
However I do know that a Faraday cage is used to shield from electromagnetic and electrostatic disturbances, and works by being electrically conductive. Faraday cages are often made of copper mesh screen, for example. What a Faraday cage will not shield is from a DC magnetic field, nor from a low frequency changing magnetic field. What is being described here is not a classic Faraday cage.
There is essentially no such thing as a magnetic insulator. There are 'diamagnetic' materials, but their ability to repel a magnetic field is pretty darn small, with the exception of superconducting materials, and these are a rather special case...superconductors resist any _change_ in the magnetic field going through them, so if you take a material and cool it down until it is a superconductor, you will 'freeze in place' any magnetic field threading through it. So when you want to shield a big magnet, the thing to do is to surround it with a magnetic _conductor_, to essentially 'short circuit' any stray magnetic flux. The description of a _steel_ shield around the room sounds quite a bit like this sort of _magnetic_ shielding.
I can understand why the company would want to strictly enforce _single_ point grounding inside this environment. Any loops of conductive material would be subject to induced voltages whenever the magnetic field changed...and if the magnet were to 'quench' then the voltages could be quite high. The magnets used for NMR imaging are in the 1 tesla range with a quite large bore area; this means that the fringe fields will be significant. (Read: grab tools out of your hand and pull them across the room, cause you to see stars if the field collapses, cause aluminium plates to fall slowly if they have to cut across field lines, etc.)
I cannot understand why they would want _no_ grounding at all; but preventing unintentional grounding strikes be as a rather good idea.
But the thing that strikes me as most wrong in this situation is apparently sending an electrician in without a very explicit design, specifying every detail of the materials to be used, the wiring layout, etc. You are the electrician, not the design engineer, and an installation that meets the requirements of the NEC is very likely not going to be sufficient here. Just what sort of conduit will you use? NEC is happy with steel conduit, but I wouldn't want it in this situation. My guess is that it is important to avoid any sort of loop area in your equipment grounding conductor system (including, for example, appliance cords plugged into receptacles at different locations, if the appliances might touch. I'm sure that there are at least 100 non standard requirements for a proper installation in this space, above and beyond the NEC requirements for patient care areas; in fact I bet some of the requirements for this space are at odds with ordinary NEC requirements (eg redundant grounding paths).
Just my two cents.
(Side note on quenching: The magnetic field in these machines is produced by a superconducting electromagnet, something which has some very interesting stability issues. Superconductors have a characteristic of zero resistance as long as their limits are respected. If the material gets too warm, it will cease to superconduct. If you try to push too much current through a superconductor, it will suddenly cease to superconduct. If you expose a superconductor to too much magnetic flux, it will cease to superconduct. What this means is that when you build a superconducting magnet, there are a number of factors which could cause a portion of the coil to simply _stop_ being a superconductor. Suddenly you will see significant resistive heating in that portion of the coil, which of course causes more of the coil to stop superconducting. The magnetic field will very quickly collapse, with all of the energy stored in that field going to heating up your coil (and any surrounding closed conductive loops). If you are lucky, when the magnet quenches all you will end up doing is boiling a bunch of _expensive_ liquid Helium. If you are not lucky, you will have significant hardware damage. I read one internet report of a 'ball lightning' being formed when a large superconducting magnet was physically damaged....)