This is not an LO bug. I get the exact same result with a Python script,
a Ruby script, and Calligra Sheets.

It is a problem with internal representation of decimal numbers in any
computer system. This is a well known problem in scientific
computing/numerical analysis.

Also, reordering the addition produced different precisions. I think the
correct numerical analysis term for this is precision referring to how
closely the actual value is to the true value.