I think you are on the right track.
TOC AND INDEX
You should be able to recover those. Check the detail of my article. Post back if no joy.
It is a while since I have done any reverse engineering so I would have to test whether or not the hierarchy is maintained. I thought it was but not sure.
It maybe there was no hierarchy in the source. You don't need one to create a TOC with hierarchy so are you sure the developers had one? I suspect they may well have ignored niceties like that. If so, you would either have to live with that or undertake it manually.
Have you considered how you will maintain this arrangement when the product is upgraded? I guess you could reverse engineer the CHM you are given and then import the topics you need, provided they don't need too much massage.
See www.grainge.org for RoboHelp and Authoring tips
From a quick scan on the Web, it appears that nDoc generates XML files first, then compiles that transitional output into HTML Help (CHM).
If the developers processed that XML output into HTML output, you could then set up links to those pages, without any need for including them as topic pages in the RH project. (Since they would be HTML files, not HTM files, you could either: 1) add them as Baggage Files for inclusion in the project; or 2) establish a folder structure for them. The developers would continue to maintain the files and the hierarchy without any effort on your part, allowing you to only work on the collateral content. You'd need to work with the developers to coordinate style sheet similarities and differences.
If you want to include those files in your search, you could look into the Zoom Search Engine, which we're already using for our large merged WebHelp output and considering for API library inclusion. Our preliminary testing has shown that these API libraries (we have a ton of them) can also be indexed and included into the Search capability. You would need to block the indexing of the API page headers and footers, for example, but the developers should be able to do that programmatically.
Thanks much for the feedback, Peter. As far as future maintenance--good question, I'm going to really have to think about how that would work.
Leon, Thanks for the feedback! The idea of linking to the html files seems like it could be a workable solution. I will investigate that further.