Small bug in calibration.py to load the reference frequency

[kazuakiyama]

I was trying calibration.py to calibrate data in alist files. I noticed that when reference frequency doesn't have a decimal point in the input alist file, it will be loaded as an array of intergers with pandas, and will cause an error in a later line below.

111. freq_orig /= (1.0e3)

I would suggest a minor fix to always load reference frequency as a float array, for instance.

99. freq_orig = np.array(df.ref_freq, dtype=np.float64)