Possible fixes to FunctionPointer PR (#28)

#28 introduced a number of questionable changes:
- the `clear` method in `FunctionPointerBase` and `FunctionPointerBind`. This was introduced because it _looks_ like it might be needed in MINAR. If it turns out that it's not, `clear` needs to be removed.
- private virtual destructor in `FunctionPointerBase` (following the already present private constructor). With that in mind, It might be that `FunctionPointerBind` and `FunctionPointerX` should not inherit from `FunctionPointerBase`, but rather include it, using composition (and a proper name change). This isn't an error, just a possible improvement if it turns out to be the right idea.

Review PR #28 and the changes involving FunctionPointerBase, FunctionPointerBind, and FunctionPointerX. Check whether MINAR requires clear, then evaluate the proposed inheritance-versus-composition design and agree on the appropriate direction. Done means the questionable changes have a settled design and are updated or removed accordingly.

Written by the indexing model from the issue text.