Preserve dtype for wrap_lons

[stephenworsley]

📰 Custom Issue

Currently, iris.analysis.cartography.wrap_lons promotes dtype to np.float64. This is causing some users to have to manually change the dtype back to its original version.

iris/lib/iris/analysis/cartography.py

Lines 70 to 73 in 98baee2

	# It is important to use 64bit floating precision when changing a floats
	# numbers range.
	lons = lons.astype(np.float64)
	return ((lons - base + period * 2) % period) + base

I believe this is done to avoid rounding errors within the calculation. If this is the case, It may be worth converting back to the original dtype after the calculation with something like the following code:

orig_dtype = lons.dtype
lons = lons.astype(np.float64)
return (((lons – base + period * 2) % period) + base).astype(orig_dtype)