Feedback on 7.0 RTTI mechanism implementation

In current master (7.0 preview), DEFINE_STANDARD_RTTI macro includes declaration and inlined definition of the static
and virtual method as follows (see Standard_Type.hxx):
static const char* get_type_name () { return #Class; } \
virtual const Handle(Standard_Type)& DynamicType() const { return STANDARD_TYPE(Class); }

At the same time previously used macro IMPLEMENT_STANDARD_RTTIEXT is defined empty (see Standard_DefineHandle.hxx).

To ensure creation of a static variable Handle_Standard_Type for each type, a method opencascade::type_instance::get() contains an explicit use of the variable myInstance what makes compiler to ensure its presence.

AFAIU, this creates the following downsides:
1. Use of inlined virtual methods would enforce the linker create duplicated implementation in each shared library.
2. Likewise for inlined static method returning a string constant.
3. No way to find a moment where the respective static variable of Handle_Standard_Type is created (as all instances are created by dynamic initializers during library loading). In some cases this can be helpful (for instance, during debugging of #26913).

Instead, why not to create (or rather restore) a more explicit and streamlined approach (used in earlier versions):
- make out of line implementations of get_type_name() and DynamicType();
- create explicit definitions of static variables defining each type.

All this can be put (back) into IMPLEMENT_STANDARD_RTTIEXT macros to be used in .cxx.

Any thoughts on this ?

Best regards,
Roman