Extensions: Change "float" parameter to use a double internally
The conversion from/to string using g_ascii_strtod / g_ascii_dtostr only works properly for doubles; for floats we produce and output with "arbitrary" numeric errors.
Edited by Patrick Storz