I think this is a good example of putting too much faith in an average measure like Impact Factor or SJR when you talk about the 'reputation' of a journal.
In 2015, in its first year, SoftwareX published "Gromacs: High performance molecular simulations through multi-level parallelism from laptops to supercomputers", which was extremely highly cited - Scopus has almost 3,800 citations, more than 40 times more than the next most-cited paper in the journal. The journal is quite small, so in 2017 SoftwareX would have had a “citations per paper” of about 8 from that one paper alone.
SJR uses a three-year window, so this paper stopped contributing to their average in 2019, which is where you see the precipitous drop in rank, as it's now reflecting the 'normal' citation rate for other papers in the journal without this huge outlier.
The question you have to answer for yourself when deciding where to submit is, should you care? You said SoftwareX had built a reputation as a respected and leading journal, so if that was actually true in your field, then nothing seems to have changed apart from this one big outlier dropping out of the citation average. On the other hand, if 'reputation' was just used as shorthand for Impact/SJR, then this has indeed substantially fallen, and you'll need to decide how much that matters to you when you decide where to submit.