Coefficients arrays must have the same dtype error in iswt function

[tallamjr]

Within iswt(coeffs, wavelet) function of pywt/pywt/_swt.py there is a dtype check for cD coefficients but not for cA coefficients.

In the following line:

dt = _check_dtype(cD)
cD = np.asarray(cD, dtype=dt)

The dtype is set for cD's but there is not setting for cA's also.

This has the effect of when running something which changes the type of cD, cA's then become of a different dtype and gives the following error:
ValueError: Coefficients arrays must have the same dtype

[grlee77]

Thanks for reporting this. I can reproduce the issue. It seems dwt/idwt does support this so it makes sense to do the same for swt/iswt as well.

I made a PR extending mixed precision coefficient support to more inverse transforms in #450.