Allow non-inplace option for ndframe.update()

[giuliobeseghi]

Code Sample, a copy-pastable example if possible

import pandas as pd

s = pd.Series([1, 2, 3, 4, 5])
t = pd.Series([10, 11, 12], index=range(3, 6))

print(s.update(t).reindex([1, 2, 3, 4, 5]))

Problem description

Obviously the previous doesn't work. s.update(t) is executed inplace, and doesn't return anything. I see no reasons why one shouldn't be able to call update and get a new object (for example to allow method chaining).

The easiest way of implementing this is by allowing a parameter that makes the function return the updated object:

print(s.update(t, inplace=False).reindex([1, 2, 3, 4, 5]))

I know there is already a plan about deprecating inplace, but I'm not sure it involves update. I guess that update was designed to be similar to the same method of a dict object.

Update
Closed as it was solved in #10730