Yes, it would be odd to have 100 mV USB. From what I can tell from looking at the insides, there is no difference from the USB on the dock and the USB on the top of the calculator, other than the dock is missing some of the protection circuit. They are separate on the inside of the calculator, so it's not the same port as the normal one.

I'm surprised there is 100 mV anything since the lowest I've ever heard ttl "high" voltages being is still above 1V. A 100mV signal would be considered "low" by most logic circuits. This would mean they are intentionally decreasing the signal voltage out of the CPU with amplifiers, which doesn't seem to be the case.

100 mV is INSANELY low for transistor logic.

In transistor logic, no, it wouldn't be. The low is determined by how low the transistors can pull (usually 0.5v to 0.7v) and is not affected by much else as far as circuits go. This limit is determined by the method by which the transistors function, and their materials. (Germanium diodes can pull down to 0.2v, for example, but they're very slow). This value is actually equivalent to the voltage drop across the diode. This means a 100 mv difference would put the "high" value at about 0.8v for standard circuits.

The high side is more arbitrary, but if you go too low (read: <1.7-ish volts for high-end consumer electronics) the circuits become unstable, reporting 0 where it should be 1, and vice versa. They push it as low as they can for speed and power reasons, as you said, but 100 mV difference is less than the thermal fluctuation, even ignoring problems like crosstalk.

My biggest reason doubt this, though, is that the CPU operates at 3.3v. 3.3v USB, while not spec, would make sense, as they would have voltage-adjusting circuitry off-board. (see the RS232 port for an example of exactly this design choice in use) Choosing to lower the voltage any more than the CPU is running, simply for I/O purposes, and taking the protocol even more out of spec? Why? It won't be faster or more efficient, thanks to conversion inefficiencies.

The nspire is not a fancy, cutting-edge piece of electronics. As far as mobile ARM devices go, it's quite ordinary, and an old ordinary device at that. I just don't see TI using such futuristic technology in a device like this.