Improve encapsulation on ERC165 and update code style guide

[ZumZoom]

Following #1282 the same prefix underscoring rules should apply to the internal state variables as to the internal functions. Luckily it is already applied in most cases.

[nventuro]

Hey @ZumZoom, thanks for bringing this up! However, I think what's really going on here is that _supportedInterfaces should be private, not internal, and it should be accessed via supportsInterface and _registerInterface. @ElOpio WDYT?

[ZumZoom]

Then it is also the case for ERC165.sol which has the same implementation.

[come-maiz]

I agree with @nventuro. We are trying to make all our state variables private. However, for ERC165 we would need to ask ourself what would this mean with relation to the ERC specification. If we make this state variable private, are we deviating from the expected interface?

[ZumZoom]

The only difference that migration from internal to private makes is impossibility to deregister interfaces. Also supportsInterface visibility is external which means child classes would use external call to access it.

From the ERC specification point of view it seems like nothing is changed.

[nventuro]

@ZumZoom you're right, but that is an unintended side effect of the current implementation: if we want to support deregistering, we should provide a method to explicitly do it. I'd go for private for the 2.0 release: we can extend this functionality in the future.

[nventuro]

Thanks for taking care of this @ZumZoom! Little fixes like this one help the whole project be more cohesive :)