I agree largely with the comments of TCHconn.

There appear to be at least two errors in the report's Table 4.1 (page 80), and these also appear in the sections excerpted by FastMover in post #5. Near the bottom, the figure for "Motor inverter specific power" improvement factor should be 1.6 and not 2.8;and the figure for "Motor inverter power density" improvement factor should be 2.1 and not 1.3.

After perusing the document, I have the following additional comments, conclusions, and conjectures:

1. The speed limit of 14 000 rpm for MG2 (and presumably also for MG1, as its construction is similar) seems to rule out mechanical over-revving of MG1 as the most likely reason why Toyota places a limit of 64 km/h (~40 miles per hour) on the vehicle before spinning up the ICE electrically when coasting in "fuel-cut" mode [pages 3 & 80].
2. Furthermore, the over-voltage (back-emf) threat to the MGs and their control electronics from high MG rpm when the car is coasting or being towed wheels-down in 'N,' also may not be the reason for Toyota's concern, since the generated back-emf, shown in Fig. 3.8 on page 61, reaches only 400 Vrms = 566 Vpeak. This is well below the claimed 650-V limit of the MGs and electronics.
3. More likely in my view is the fact that the transaxle's primary lubrication comes from the oil pump driven from the ICE's output shaft [pages 39 & 40]. If the ICE was not spun (in order to operate the transaxle oil pump) while coasting at speed, there would probably be insufficient lubrication of the transaxle's gears and bearings. The oil pump drive cannot simply be moved to the MG2 output shaft instead, since then there would be no lubrication if the ICE were running with the car stationary — for example when the ICE is charging the battery, warming up, or providing heat for the cabin. So, the oil pump is driven by the appropriate shaft. [It now appears that I was mistaken when I said in another thread that the transaxle oil pump was driven by the MG2 output shaft. I shall correct that error.]
4. Pages 65 - 71, and especially Figs. 3.18, 3.20, & 3.22 show that the combined (motor+inverter) efficiency is >90% over much of the operating region in the rpm-torque plane, and even reaches as high as 92% in the central region. So, my initial assumption of 90% efficiency (post #1) in the thread "An Unpalatable Fact (with apologies to Al Gore!)" was reasonably close, and my post-experiment deduction (posts #58 & 83) that it could in fact be significantly >90% wasn't far out either.

Stan