Optimize _PyObject_IS_GC in free-threaded build

[colesbury]

Feature or enhancement

@nascheme writes the following suggestion (from #114564 (comment)):

_PyObject_IS_GC() could be optimized in the nogil build. That function is a bit expensive since it has to check for tp_is_gc() and possibly call it. Now that all objects have ob_gc_bits, we could store the "IS GC" status as another bit there, when the object is allocated.

I think this might require a bit of refactoring, but worth doing eventually. At a glance:

• _PyObject_GC_New() (and related functions) could initialize ob_gc_bits with the "IS GC" set
• PyType_GenericAlloc() would also need to initialize ob_gc_bits with the "IS GC" set if the type is a GC type.
• _Py_NewReference() requires some care. It. currently initializes ob_gc_bits to zero and it's used after "allocating" objects from freelists.