Since a thermal printer only needs as much pressure on the platen roller as required to keep the paper registered, the roller itself could be hollow and thus carry negligible inertia. The print head could be net neutral by operating boustrophedonically, which many printers already do - both of my inkjets, for example! - and that might actually count for more, considering the possible need for some thermal mass to dissipate waste heat from the printing elements. (I don't think modern thermal printers need that, but I suppose in the 80s they might have.) Worst case, the firmware might need to ensure the number of head passes is even, I think most simply by counting print passes and if needed performing a final "return" pass with a cold head - again a common behavior, and one also displayed by both my inkjets since that avoids the need for a parking station on either side of the paper path.
The line printer with its heavy drum would probably be more of a concern for the Shuttle, although I doubt a significant one. Reaction wheels function by change in angular momentum, and a line printer drum spins at a fixed rate in operation; as long as spinup and spindown take about the same time, which is easy enough to achieve, the net effect on attitude should be negligible, especially considering the entire printer constitutes only about 1/3000 of the orbiter's dry mass.
You might not want to run the thing during an OMS burn, but even if you did I doubt it'd matter; assuming constant thrust from spinup through spindown, your course would be displaced proportionally to the length of the print job, but given the minimal relative inertia and that line printers typically output in the tens of hundreds of lines per second I think it'd need quite a long job to make a noticeable difference, and you'd almost certainly run out of paper long before.
(Don't take any of this too seriously; I'm the wrong kind of engineer to give authoritative answers here, but the question is fun to think about anyway.)
