A few background facts:
1. Common-mode voltage is, by definition, the average of the two voltages (with respect to ground) on the two signal conductors (pin 2 and 3 of XLR);
2. Impedance balancing of driver, line, and receiver produces all the noise rejection benefits of a balanced interface (and that's why we use them);
3. Signal symmetry (equal and opposite signal swings on the two lines) has some benefits but they relate only to headroom and crosstalk issues, not noise rejection;
4. A symmetric balanced line produces no common-mode voltage when signal is present, but a non-symmetric balanced line produces a common-mode voltage equal to half the signal level.
5. Amplifiers with balanced inputs will have two important characteristics that limit their performance: a)common-mode rejection ratio or CMRR (the ratio, in dB, of its response to a differential signal input compared to its response to the same signal level applied as a common-mode voltage) and b) common-mode voltage limits (this varies widely depending on design and, sadly, is rarely stated in specifications). For amplifiers with good transformer inputs, CMRR in the 90 to 120 dB range is easily achieved and common-mode voltage limits are those of the transformer's internal insulation, generally several hundred volts at least.
As best I can determine the actual schematic of the Folcrom RMS 216, although it attenuates the (differential) signal by some 30 or 40 dB, it attenuates the common-mode voltage very little. Therefore, if a symmetric balanced output drives an input, the signal is attenuated and little common-mode voltage is produced. However, if a non-symmetric balanced output drives an input, a common-mode voltage will be produced at its output that is much larger than the differential signal. This poses two problems for the input driven by the Folcrom RMS 216 output. First, it will require high CMRR to reject common-mode that is potentially some 30 dB larger than the signal itself ... likely only a transformer input will have enough CMRR. Second, the high common-mode voltage may exceed its limits ... this generally results in particularly ugly distortion with a sudden onset as signal level is increased. These problems are created by the circuit design of the Folcrom RMS 216. Since input characteristics of the JCF model LEVR are so sketchy, I don't think we can assume it has either high CMRR or a large common-mode voltage range ... but the JT-MB-CA will add both.
A conventional active summing device would not have such problems and, as an added benefit, would not need to be followed by makeup gain. I think the perceived benefit of a "passive" summing network is lost if it must be followed by an "active" amplifier. Why not use the amplifier in the summing network? You still only have one amplifier in the signal chain, but none of the problems we're discussing.
