Passing XML as the parameter is actually another good use-case for a similar technique, and I've done this where I was dealing with an object-level interface. In that way, the parameter can be passed an entire object worth of data, or for that matter, a set of objects, which can have parent-child relationships encoded within the XML (I used this to pass sets of invoices around between the data layer in the middle tier and the database).
However, that's not where I have used the technique of comma-delimited strings as parameters; that's more about working with reporting services. I've scratched my head a bit, but I can't think of another case where I've passed sets of values as a delimited string. I have seen it used to express lists of values in a "parameter" table, but I think that's just silliness and not a valid reason to use this technique (yes, we are allowed to have tables with relationships!)
I will admit that once I am in T-SQL, if I have to pass sets around as parameters, I will do it either within a table or as a table-variable. If I only ever lived in SQL Server (or Oracle), I wouldn't need to be passing sets as delimited strings.
I wonder what TRON would have looked like if it were based on an RDBMS rather than a game?
My other thought was that SQL as a language is based around working with sets of data. It shouldn't matter too much about how that data is encoded; whether it's XML, a delimited string, a file, a spread sheet or a table within the database, I still want to be able to apply the same capabilities of SQL to it. From that point of view, taking a comma-delimited string and expressing it as a table is perfectly valid.