Source analysis needs to catch missing commas in API calls

[hostilefork]

If you write:

rebElide("catch ["
    "print {asdf}"
    "print {jkl;}"
"]");

C will merge those strings without spaces, so you get print {asdf}print{jkl;}. If you use commas in the variadic API the way you're supposed to, it will work correctly.

It's an easy mistake to make, especially because it works sometimes. e.g. the catch [print is legal to not have a space.

Source analysis could catch these cases. It also might be nice if __FILE__ and __LINE__ information could be injected by some preprocessing step so the scanner could have that data:

rebElide(rebWhereFile(__FILE__), rebWhereLine(__LINE__), "catch [",
    rebWhereLine(__LINE__), "print {asdf}",
    rebWhereLine(__LINE__), "print {jkl;}"
rebWhereLine(__LINE__), "]");

This would be optional, but very useful.