Apologies for the English and for the long response, but hopefully this clears up some of the doubts.
For various reasons, the OHM project is in the public domain “unless otherwise noted”, quite unlike OSM. The exceptions still need to be available under open licenses, of course. We don’t allow contributions that would cause the entirety of OHM to become subject to a particular license, since that could conflict with other licenses and turn “unless otherwise noted” into quite an understatement. 
The ODbL applies to any derivative databases, which precludes a wholesale import of ODbL-licensed OSM data.
That said, it is technically possible to bring in an “insubstantial” amount of ODbL data without triggering many of the ODbL’s provisions. The OSMF has issued guidance on what counts as insubstantial, which is wonderful, because IANAL. Based on this guidance, are the abandoned railways in these four changesets insubstantial? Probably, so I guess transferring them would be OK. But please don’t designate someone to serve as the “OHM transfer specialist” to be called upon repeatedly any time someone needs to transfer railways anywhere in Germany. That would probably be much riskier.
Admittedly, this is all very complicated. I don’t think the OSMF really wants the ODbL to be an obstacle as this community figures out how to handle these historical features. The people responsible for OSM and OHM 
probably need to get together at some point to discuss options for simplifying this situation.
OHM also has some expectations besides legalities. We really, really like it when mappers provide start and end dates, even if it’s just an educated guess. Otherwise, we assume the feature has always existed. If you find these abandoned railways annoying in OSM, imagine how annoying it would be to see standard-gauge railways crisscrossing the Carolingian Empire! Fortunately, this mapper seems to know some of the end dates. Maybe they know the start dates too, if we ask?
We also really, really like it when mappers explain how they know. Did these changesets really come from “local knowledge” alone? All these rails, switches, and dates, with such precision? Maybe so, but if not, we would welcome more detailed information. Many OSM contributors are unaware that unoriginal facts are ineligible for copyright protection, so they aren’t as forthcoming as they should be about their sources. In OHM, mappers routinely, shamelessly rely on copyrighted materials to learn facts, as long as they cite their sources. It can’t hurt to ask. And of course, OSM would also be a “source” in this case.
Given these other expectations, sometimes it’s just easier to recreate the features from scratch, especially if you can infer them from the same aerial imagery that the original mapper used. Unfortunately, in this case, the rails aren’t visible in imagery, so researching the features would be more tedious.
The wiki documents the procedure for transferring a feature to OHM. The warning there recommends asking for permission before transferring another mapper’s contributions, as a best practice. The reason is that, as an OSM contributor, you retain ownership of your contributions and can also allow them to be contributed to other projects under different terms, such as CC0. Getting the original contributor’s permission makes things easier for everyone. You wouldn’t need to read half of what I just wrote.
If anyone here would like to discuss OHM on an open platform, there’s also the OHM Forum, which accepts your OHM account. Discourse Chat is enabled too. To see the chat room, you need to first follow Discourse’s introductory tutorial. We have some German speakers on the forum, so you don’t have to read so much of my English.
und hänge mich hier, dankenswerterweise, gleich mal mit an.
) , man kommt aber auch über Matrix rein, wenns wär.