True if using the fork sc-im, but unfortunately not quite so true for the original sc.

Similarly, the venerable GNU Oleo docs say it can import from CSV directly, while the active fork Neoleo recommends doing this via provided Python script:
Well, theoretically it can import from CSV directly, too, but practically, this works only with simplest CSV data. And I believe the original Oleo would have had problems with complex CSV data as well, I just haven't tried it out in ages.

Also, be aware that WWW::Splunk as packaged on some distributions may include a competely unrelated /usr/bin/sc (Splunk Client).

For the sake of completeness, let's also mention tabview, although their own README recommends using VisiData for more complex tasks.

VisiData is good for when the data is cleaned and ready to be analyzed.

You can clean up data from inside VisiData, too.

Open Refine is my go-to program for cleaning data when code, for whatever reason, can't cut it.

It gets more complex. CSV isn't really well-defined and in practice such files can be full of a lot of weirdness.

However, for files generated in-house, you'd have control over how the file is set up and what is used to mark fields and records. Then it would not be a hard problem.