The way RM handles individual facts such as birth, death, and burial in reports vs. the way it handles couple facts such as marriage and divorce is more or less standard in narrative reports for genealogy. That doesn't mean it's always the preferred way that it be done, but RM isn't way out in left field.
As for myself, I like to have marriage and divorce facts listed in timeline order in narrative reports. Therefore I use RM's shared fact facility to share the marriage and divorce facts with the principals to the marriage and divorce. This may sound strange because the marriage and divorce facts are already associated with the couple. But it has a salutary effect on my narrative reports.
RM's narrative reports have an "individual section" for each person and a "family section" each person. In the NGSQ format (modified register format) and in the NEHGS format (register format) the "family section" of the report includes family facts such as marriage and divorce in addition to a list of children. The inclusion of the marriage and divorce facts in the "family section" along with me forcing the same facts into the "individual section" results in a duplication of the facts. I use white space to separate the "individual section" from the "family section" so the effect of the duplication is not actually so bad. Also, I include a lot more information about the marriage and divorce in the "family section" than I do in in the "individual section".
For the purposes of composing this message, I tried telling RM not to include the marriage fact in narrative reports in hopes that it might suppress the main marriage fact in the "family section" of the report while still displaying the shared version of the marriage fact in the "individual section" of the report. But alas, doing so suppressed the marriage fact both places. Essentially, RM supports lots of options for when to include or not include facts, but the roles for those facts don't have such options. Instead, roles inherit their options from their respective owning facts. I hadn't really thought about this before, but if RM's shared facts are to be a really robust and useful feature of the product moving forward, then roles need all the same include/exclude options as do the main facts.