If the isolation happens outisde the metal case well you do not need ground it. It that case yes connect it to 0VDC for EMC stuff.
Yes, this is what I do with e.g. 5V USB microcontrollers in die-cast aluminium enclosures, as there is no separate "ground", only USB 5V and USB GND. (That is, the USB GND is connected to the chassis at or very near the USB connector. But I do wonder if I should put a big-ish resistor in between, to absorb the energy from static discharges from humans to the enclosure.)
This is a topic that might make a good Youtube video for someone like Dave or any of you EEs here. Me no electrichicken, just a unclebumble.
Most current stepper board controller boards (all with integrated or plug-in stepper drivers) use a shared ground for motors and logic. This is problematic, because if connecting via USB, you really need an USB isolator: the motor supply is separate from the computer power supply, and may not be at the same potential (or phase; consider electrical mains derived from two different legs of three phase power source). Also, the motors do generate noise, which is not that easy to fully filter out from the logic side, because the logic stuff consumes a minuscule amount of current (50mA - 150mA at 5V) compared to the steppers. The boards I've seen from RAMPS to SmoothieBoard just use several filter caps on the logic supply input and output pins and the MCU (and other external active IC) supply pins, use a shared motor and logic ground, and tell users to use the device "offline" (transferring files via memory cards or USB memory sticks), or have an Ethernet or WiFi connection for online use.
These 3D printers and mini-CNCs have such a board, a metal support frame, and a 12V or 24V AC-DC power supply (mainly for the steppers, but usually also used for the logic via a separate DC-DC converter). If the AC supply has a separate ground connection (non-double-insulated supply), it should be connected to the frame; if it only has two AC wires, then the DC 0V should be connected to the frame – but should there be a power resistor or capacitor in series or in parallel to absorb static shocks, say someone coiling PLA in dry conditions with just the wrong gloves on, and accumulating a rather high static charge, and discharging it through the metal frame?
(Those who have a whiteboard in their office: I recommend you put a big carbon resistor, say 1M, between the aluminium pen tray and a nearby grounding point. No more annoying zings from static discharge. Should one put a capacitor in parallel to the resistor also? Why?)
So, there is a point on A) how to design better (or a simple example of using external bigger drivers (up to several amps peak per phase) with optoisolators); B) how to safely connect an "offline" printer with shared motor-logic supply to a PC via serial or USB-serial using an USB isolator (the el-cheapo Chinese clones of ADuM3160 appnote, that work with low and full speed USB 2.0 (1Mbit/s and 12Mbit/s respectively), have worked perfectly for me thus far); and C) how us hobbyists building our own printer enclosures and tinkering with our printers etc. should deal with grounding and static shock dissipation sensibly.