Yes, a little Latin can be useful, especially in the older records -- the priests did like to use it, not just in the kind of "leared names" that we are discussing here but also in dates, in causes of death and sometimes just to record things that they didn't want others to be able to read. But as for these kinds of names, it is generally enough to know that they tend to end in -us for men and -a for women, and that the genitive forms that are used for patronymics instead end in -i and -ae, respectively (giving names that end in -ius, like Celsius, the distinctive -ii ending).
Once you recognize a name as being Latin, the next step is decoding it. First names tend to be Latinized forms of Swedish names -- sometimes these are actual Latin names, such as Nicolaus and Magnus (standing for the Swedish Nils and Måns, respectively, which were historically formed from those Latin names), but in other cases they are newly formed, such as Olaus for Olov (a Germanic name that never had a counterpart in classical Latin), simply to allow the name to fit into one of the Latin declinations so that it could be used in Latin texts where it would need to recieve case endings. This is something every priest did as they studied Latin, since they needed to be able to talk and write about themselves and each other. Similarly with patronymics, which as described above are simply genitive forms.
Just as with the later soldier names, there were often several people with the same name and patronymic studying at the same time. To make it possible to distinguish people within such groups of students, those who didn't have an inherited family name also tended to adopt or acquire a second name; these were often based on where they were born or lived before they went to university, but could also later in life be based on where they worked. The name Alptaneus that I mentioned in my last post was born by the father of Magnus Celsius, is formed to the parish name Alfta where he worked before he became a vicar; another example which is still in use today is Moraeus, from the town of Mora. Some of these names were inherited and became family names, while others were not. These names can be useful in that they tell where people came from, and they shouldn't really be much of a problem -- at least not compared to keeping track of the Latinized and non-Latized forms of the first name and patronymic.
These names are not really different names from their Swedish counterparts, they are different forms of the same name and were used at the same time depending on the situation and who was talking to whom. And of course we don't know to which extent each form was actually in use, we only know what the priest writing the record chose to use. Which brings us to your question, finally! (Sorry for the wall of text...)
Yes, I too understand WikiTree policy is to use the birth/baptism record (if it can be found). This is the record that is most likely to be accurate, in the sense that it alone will not contain any copy errors (a great-grandaunt of mine was named Eterna, but after moving around a bit the forms written in the household records look more like Elvira).
However, when it comes to different forms of the same name, especially in old records before spelling was standardized, there is no guarantee that the birth record is more correct than any other record made during the person's life, because the choice between forms may have been made by the recording priest just as well as by the parents (or later in life the person themselves). On the other hand, neither is it worse than any other record, and we do need some kind of standard, so we might as well use that form. Unless, of course, there is a project style guide that says otherwise; style guides always overrule any general rule. (And in the specific case of Nils Celsius it seems we don't have a birth record, which gives us a bit more leeway. But again, I could be quite wrong in my admittedly not very deeply considered opinion; there ought to be a style guide about this somewhere, created by people with more experience than I have...)
Also, note that Swedish birth records (at least historically) generally don't explicitly state the child's last name, which sometimes makes the LNAB question less obvious than one would expect.
Using "Preferred first name" to present a different form of the "Proper first name" seems wrong to me, though, partly because we again don't know which form was actually preferred, but mostly because I think the point of the "Preferred first name" field is to show that a middle name should be used instead of the "Proper first name", and selection between names is different than selection between forms of names. (As an example: my full name is Jerker Olov Georg Winstrand -- "Proper first name" is the first in the list, but the one I actually go by is Olov, so that goes into "Preferred first name".)
