gh-87613: Use AC vectorcall for float (GH-158201)

Move float.__new__ to AC vectorcall dropping the hand-written.

Running microbenchmark from the original add (`pyperf timeit "float(0)"`
in gh-86039) shows no significant performance change.

The generated name matches the hand-written name so the tp_vectorcall
slot is unchanged.

Update the vectorcall slot-parity test to include float.

Author: cmaloney

Lib/test/test_clinic.py

@@ -5665,7 +5665,8 @@ def through_init(cls):
             return cls, partial(cls.__init__, cls(1))
 
         entry_points = [
-            through_new(enumerate),   # the only non-test @vectorcall function
+            through_new(enumerate),
+            through_new(float),
             through_new(ac_tester.VcNew),
             through_new(ac_tester.VcNewBase),
             through_new(ac_tester.VcKwOnly),

Objects/clinic/floatobject.c.h

@@ -1556,6 +1556,7 @@ static PyObject *
 float_subtype_new(PyTypeObject *type, PyObject *x);
 
 /*[clinic input]
+@vectorcall
 @classmethod
 float.__new__ as float_new
     x: object(c_default="NULL") = 0
@@ -1566,7 +1567,7 @@ Convert a string or number to a floating-point number, if possible.
 
 static PyObject *
 float_new_impl(PyTypeObject *type, PyObject *x)
-/*[clinic end generated code: output=ccf1e8dc460ba6ba input=55909f888aa0c8a6]*/
+/*[clinic end generated code: output=ccf1e8dc460ba6ba input=39b40d663c372a91]*/
 {
     if (type != &PyFloat_Type) {
         if (x == NULL) {
@@ -1610,23 +1611,6 @@ float_subtype_new(PyTypeObject *type, PyObject *x)
     return newobj;
 }
 
-static PyObject *
-float_vectorcall(PyObject *type, PyObject *const *args,
-                 size_t nargsf, PyObject *kwnames)
-{
-    if (!_PyArg_NoKwnames("float", kwnames)) {
-        return NULL;
-    }
-
-    Py_ssize_t nargs = PyVectorcall_NARGS(nargsf);
-    if (!_PyArg_CheckPositional("float", nargs, 0, 1)) {
-        return NULL;
-    }
-
-    PyObject *x = nargs >= 1 ? args[0] : NULL;
-    return float_new_impl(_PyType_CAST(type), x);
-}
-
 
 /*[clinic input]
 @classmethod