It is not necessary and not efficient to use a std::vector for max. 8 elements only to iterate them once for calculating the sum of the input: The sum can be calculated directly instead.

Originally, this spot caught my attention because

cen = Vector_2(cen.x()/EScalar(hits.size()),cen.y()/EScalar(hits.size()));

caused issues in x86 MSVC.

Unfortunately, it is not covered by a test. Offering it as a draft.

Checklist

• All changes meet libigl style-guidelines.
• Adds new .cpp file.
• Adds corresponding unit test.
• This is a minor change.