tl;dr: They could be, but if the question specifically asks about a specific software package (e.g., FEniCS), it's probably going to get closed with a reply to post the question on a mailing list devoted to that package.
The tricky thing with tool questions mainly has to do with the level of familiarity needed with the tool, and the level of detail required to answer the question.
What we've discovered over 4 years of running the site is that very package-specific questions (e.g., How do I create a block preconditioner in PETSc? or How would I implement Nitsche's penalty method in FEniCS?) are best directed to package-specific mailing lists. In most cases, there aren't enough power users or developers of these packages to answer these types questions here, and they tend to prefer answering these questions all in one place, on their mailing lists. Also, if the package has a larger, ardent user base, these questions can swamp out more general computational science questions on the site, as we found out with FEniCS, and if the package has a smaller user base, these questions can get lost in all of the other questions, as we found out with CVX.
We've also discovered that recommendations on packages (e.g., Are there any packages that implement an incremental SVD? If I want to solve a semi-definite program, what package should I use?) work out well because one doesn't have to be a very experienced user or developer of these packages to provide good answers. These are the sorts of questions that are a better fit for us, because they usually aren't asked on mailing lists, since the answers tend to involve multiple software recommendations.
So what I'm saying is, if someone asks a question about data movement, the answer may include a recommendation to use Globus, but if someone says, "How do I distribute my data using Globus?", a moderator or user will politely close the question and suggest asking the question on the Globus mailing list.