As I wrote in another post, those graduate students were at universities and thus also had teaching duties in addition to their research positions. While the vast majority of those grad students vacated their teaching positions during the strike, many continued with their research. Academic strikes work well at teaching institutions, where the loss of instructor labor is hugely consequential to the university employer (and has no impact on the striking instructors' careers), but not at pure research institutes like the NIH.

Isn't this something trivially resolved by the fact that the union members will be the researchers who can decide how to organize themselves? If they feel that their individual research priorities are more important than organizing for greater protections against sexual harassment or whatever-- that's something that can be voted on!

No, this is not a trivial problem, it is more like a prisoner's dilemma. A great majority of union members can agree that there should be a strike in favor of greater protections, remuneration, or benefits, but it is still in each of their individual interests to continue to work for their own benefits. Put differently, they each want the collective to strike, while continuing their own work. I suspect that a 'strike' would involve most/all of the members delaying submission of findings, but continuing any work that was 'low visibility' (i.e. didn't require outside resources).

I mean, the article explicitly states that as federal employees they may not strike. So I presume there are other things they will plan, such as public demonstrations.