Don't apologize for butting in - the more butts the better! You ask a lot of good questions that demand good answers.
You are right about HSPA and voice, the service is carried over circuit-switched dedicated channels for which CDMA is an excellent solution in terms of QoS and capacity. However, this is not a fall back scenario from HSPA, it is very much the designed parallel operation that caries none of the drawbacks associated with CSFB from LTE to 2G or 3G.
You ask why a voice solution matters to LTE when it is primarily a data service. The reason needs to be looked at from the other direction. For an LTE-enabled smart phone on an LTE connection, an incoming voice page will cause the LTE link to ungraciously drop to enable the voice call to be picked up on 2G or 3G after some delay. The state of the LTE connection and any apps is presumably left indeterminate. For a data-only dongle the problem is much less.
So LTE may not be targetting voice, but withouth voice support, the voice service will destroy the LTE link. That is a new problem for which the industry has yet to agree one solution. I think there are around 6 on the table and no silver bullets. The prospect is further fragmentation.
Dual transceivers could help but considerably drive up costs.
Your points about the longevity of 2G and 3G are well taken and I agree it will be a long time before legacy networks are turned off, particualrly when there is still active evelopment on EDGE and HSPA.
What remains to be proven is whether deployment constraints - primarily due to spectrum limitations - will enable LTE to emerge as a significantly higher efficiency or higher capacity air interface. This is a complex subject which has no single answer as it depends on the operator's particular footprint. One critical element of this is that LTE's efficiency in the same bandwidth is similar to HSPA+ (within 20% on paper) so the big difference comes with spectrum.
If an operator does not have new spectrum, switching to LTE would not look good. If new spectrum is available, the choice of technology is still not clear since the economies of scale of extending an HSPA network, e.g. AT&T's stated plans, has to be compared to the cost of operating a new network and dealing with the fragmentation and interworking issues. The dynamics for a new operator with no legacy network are different again but worldwide scale matters here if device costs are to be kept low, and multi-format multi-band 2G + 3G + 4G devices will be anything but cheap for a very long time.