Error thrown when trying to set charge to None of an Atom with charge None and Coulombic interaction.

**Description of the error**
Upon creating an *Atom* object and initialising a Coulombic interaction for it without specifying the charge:

``H1 = Atom('H')``
``Coulombic(atoms=H1)``

The *Atom* object now has a charge of ``None`` and a Coulombic interaction. Currently, there is implementation in MDMC so that an error is thrown whenever there is an attempt to set the charge of the *Atom* to ``None`` when there exists a Coulombic interaction for that *Atom* object.

When trying to set the charge of *Atom* to ``None`` (when it's charge is already ``None``), this error is thrown:

``H1.charge = None``
``ValueError: Can't set charge to None when a Coulombic interaction exists.``

**Describe the expected result**
This problem arises as a result of very unexpected user behaviour, hence the low priority. However, an error should not be thrown in this particular case; initialising a Coulombic interaction with a charge of ``None`` does not create an associated interaction function so should not throw an error when the charge is set again to ``None``.

Setting the charge to ``None`` when it is already ``None`` should result in no change of the properties of *Atom*; it's charge should still be ``None`` and it should still have an initialised Coulombic interaction, but an interaction function should also not be created in the process.

**Suggested fix**
There should be a mechanism by which an error is thrown **only** when trying to set the charge of an *Atom* to ``None`` when it has a Coulombic interaction **and** an associated interaction (i.e. when it's charge is a ``float`` value).

**Additional Information**
This issue is linked to issue #212, so that the expected behaviour in both cases is as follows.
When the *Atom*'s charge is ``None`` and it has a Coulombic interaction but no interaction function:
* setting the charge to ``None`` shouldn't throw an error nor should it create an interaction function. The charge should remain as ``None`` and the *Atom* should still have a Coulombic interaction.
* setting the charge to a ``float`` value should create an interaction function associated to the Coulombic interaction.