Thank you for your kind comments.
As far as my RM data being able to be converted to any other software platform, I share your concerns. The way I have come to think about the problem, the best approach is to separate data from metadata. That means that I have an expectation that any other software platform will retain things like facts and notes and sources. But that also means that I don't have any particular expectation that any other software platform will retain things like my sentence templates and my source templates. Well, in the case of my source templates, the loss of the templates in another software platform is ok as long as the citation sentences themselves are preserved, and this is pretty much the case with my data.
I agree that not being able to provide sources for notes is a serious problem in RM. There is not a really good solution. The only thing I've done that sort of tends in that direction is that the point form report that I first posted in this thread has since evolved even further. I used to include transcripts of death certificates and obituaries in the note for RM's death fact. I now have three "death facts" - the standard death fact, a death certificate fact, and an obituary fact. The death certificate fact and the obituary fact are essentially facts which only have a note. So an obituary itself serves as evidence (a source - a citation - a source citation - or whatever) for the obituary fact and for the death fact. A death certificate itself serves as evidence for the death certificate fact and for the death fact. This approach lets me be slightly more targeted with my sourcing, and it also lets me if I wish to produce reports that don't include obituary transcriptions or death certificate transcriptions. I'm doing the same thing with several other kinds of facts where I break the fact into component parts be creating new user defined facts (birth certificate fact, birth notice in the newspaper fact, tombstone transcription fact, tombstone GPS fact, etc.). Even though I say I'm being more targeted with my sourcing for notes such as obituaries and death certificates, the citation superscript comes before the note instead of as a part of the note or after it. There is really nothing I can do in RM to correct this situation.
Well, there kind of is a way in RM I could have gotten the citation superscript to the end of a note, but it's so obnoxious that I didn't want to go there and it would have caused data not to be shared well with other software platforms. Namely, I could have created a "citation only" fact where the sentence template was completely null. I could have put such a fact after the note and the citation would have been in the correct place. I got the idea from RM's direct import of TMG data. TMG has a way to provide evidence for parentage, e.g., a place in the data to which such evidence can be attached. RM has no such place for evidence of parentage, except in the sense that you can attach such evidence to the "general person" data for a person. So when RM imports evidence of parentage from TMG into RM, RM creates sort of a dummy fact called ChildParent to which the evidence of parentage is attached. The ChildParent fact which RM creates has a null sentence template. That way, the evidence of parentage is not lost on import from TMG to RM, and there is even a citation superscript when RM prints a report. The only problem is that the citation superscript is sort of dangling in midair since it has no sentence to which it is connected.
Back to RM data and other software platforms, as a former TMG user you are surely familiar with a software platform called SecondSite which makes Web pages from TMG data and with a new product called GedSite. GedSite is very much like SecondSite except that it makes Web pages from GEDCOM and the GEDCOM can come from any software platform that can make GEDCOM. (Disclaimer: GedSite is not an RM product and might be viewed as a competitor to the Web pages that RM itself can make.) GedSite is very clever in reading GEDCOM from other software platforms in that if said GEDCOM includes metadata such as sentence templates, then GedSite will take advantage of such metadata and format Web pages appropriately. GedSite does an excellent job of interpreting RM's GEDCOM and in formatting RM's data based on RM's sentence templates. But I have nevertheless found it very advantageous to ignore that capability of GedSite and instead to enter metadata such as sentence templates and date templates directly into GedSite and to tell GedSite to ignore RM's own metadata. This is very consistent with my notion of separating data from metadata as way of not having to worry about my RM data in other platforms.