[WIP] Improve user facing API: make Object ctor from Obj* private

[vvolkl]

This PR makes this and similar constructors

edm4hep::MCParticle::MCParticle  (  MCParticleObj *  obj  ) 

private. As far as I can see there is no way for users to access a MCParticleObj* and they are not intended to do so, so this should not be part of the public API.

The downside implementationwise is that the constructors templating macro does not cover private, so I had to move it out of the macro.

BEGINRELEASENOTES

• Improve user facing API: make Object ctor from Obj* private

ENDRELEASENOTES

[tmadlener]

I agree that the constructor from an Obj* should be private. However, it is not entirely trivial, because it is also used for OneToOneRelations that are missing or not (yet) set, i.e. the constructor from a nullptr below.

podio/python/templates/macros/implementations.jinja2

Lines 79 to 84 in ed207e8

	const {{ relation.relation_type }} {{ class_type }}::{{ relation.getter_name(get_syntax) }}() const {
	if (!m_obj->m_{{ relation.name }}) {
	return {{ relation.relation_type }}(nullptr);
	}
	return {{ relation.relation_type }}(*(m_obj->m_{{ relation.name }}));
	}

One way out of this would be to declare all of the related classes as friend, but that can become quite a list, and I am not sure we want to go into that direction.

[tmadlener]

I just realized this constructor is also used in setReferences, e.g. here:

https://github.com/AIDASoft/podio/blob/master/python/templates/macros/collections.jinja2#L63-L69

[vvolkl]

Ok, first step is then probably to update to doc string to clearly state that this is an internal method, and then fix this in the long run.

[tmadlener]

Superseded by #514