Eliminate overhead of fetching and testing `NULL` attributes in `STORE_ATTR` specializations for new objects

[markshannon]

Consider

class C:
    def __init__(self, a, b, c):
        self.a = a
        self.b = b
        self.c = c
C(1,2,3)

This produces a trace looking something like this:

...
_CHECK_AND_ALLOCATE_OBJECT
_CREATE_INIT_FRAME
_PUSH_FRAME
# Some guards
_LOAD_FAST_BORROW_1
_LOAD_FAST_BORROW_0
# Some more guards
_STORE_ATTR_INSTANCE_VALUE
# Some more guards
_LOAD_FAST_BORROW_2
_LOAD_FAST_BORROW_0
# Some more guards
_STORE_ATTR_INSTANCE_VALUE
# Some more guards
_LOAD_FAST_BORROW_3
_LOAD_FAST_BORROW_0
# Some more guards
_STORE_ATTR_INSTANCE_VALUE
...

Each of those _STORE_ATTR_INSTANCE_VALUE reads the old value out of memory and then conditionally decrefs it.
But in this case we know that the old value was NULL so we can just overwrite it.
So we can replace this:

        PyObject **value_ptr = (PyObject**)(((char *)owner_o) + offset);
        PyObject *old_value = *value_ptr;
        FT_ATOMIC_STORE_PTR_RELEASE(*value_ptr, PyStackRef_AsPyObjectSteal(value));
        if (old_value == NULL) {
            PyDictValues *values = _PyObject_InlineValues(owner_o);
            Py_ssize_t index = value_ptr - values->values;
            _PyDictValues_AddToInsertionOrder(values, index);
        }
        Py_XDECREF(old_value);

with this:

        PyObject **value_ptr = (PyObject**)(((char *)owner_o) + offset);
        FT_ATOMIC_STORE_PTR_RELEASE(*value_ptr, PyStackRef_AsPyObjectSteal(value));
        PyDictValues *values = _PyObject_InlineValues(owner_o);
        Py_ssize_t index = value_ptr - values->values;
       _PyDictValues_AddToInsertionOrder(values, index);

On Aarch64, this reduces the number of machine instructions from 48 to 26.

The same reasoning also applies to _STORE_ATTR_SLOT where it reduces the number of machine instructions from 32 to 14.

See also #134584

We can probably remove some of those guards as well, but that's a separate issue.

[markshannon]

As a mechanism for achieving this we can add a new jit optimizer type that tracks the NULLness of its attributes.
It would need to track when it was last modified, to check for escapes, and one bit of information for each slot.

    uint8_t tag;
    uint8_t slot_state[9];  // Handles up to 72 slots, which should be plenty for most classes.
    uint16_t last_modified_uop;
    uint32_t type_version;  // Can use this to remove some of those guards

[cocolato]

If needed, I can try to implement it.

[Fidget-Spinner]

@cocolato your __slots__ PR can already do this. It needs a bit of adjusting to track escapes, but tracking the types of something in slots is a superset of tracking the nullness of the slot.

@markshannon let's use #144122 to achieve this. I already discussed with @cocolato on the issue that we should do this. It's a useful thing to have.