Draft: Default EXTXYZ writer to use 16 decimal places

This change makes it so that the EXTXYZ writer defaults to using 16 decimal places instead of 8 when writing floats. The formatting code relies upon the use of dtype.kind, which returns the string 'f' for both numpy.float32 and numpy.float64. This caused the format string to write using only 8 decimal places, which would mean that an EXTXYZ file that had (for example) atomic positions written to 16 decimal places would return False when comparing an atoms object that was read in from EXTXYZ, written back out to EXTXYZ, then re-read from the new file.