v2.3.2

Update from real*10 to modern floating point precision specification

Now uses selected_real_kind() to choose an appropriate, supported precision. Allows compiling (but not successful running) on Apple Silicon.