Issue #12911: Fix memory consumption when calculating the repr() of huge tuples or lists.

This introduces a small private API for this common pattern.
The issue has been discovered thanks to Martin's huge-mem buildbot.

Include/Python.h

@@ -101,7 +101,7 @@
 #include "warnings.h"
 #include "weakrefobject.h"
 #include "structseq.h"
-
+#include "accu.h"
 
 #include "codecs.h"
 #include "pyerrors.h"

Include/accu.h

@@ -0,0 +1,35 @@
+#ifndef Py_LIMITED_API
+#ifndef Py_ACCU_H
+#define Py_ACCU_H
+
+/*** This is a private API for use by the interpreter and the stdlib.
+ *** Its definition may be changed or removed at any moment.
+ ***/
+
+/*
+ * A two-level accumulator of unicode objects that avoids both the overhead
+ * of keeping a huge number of small separate objects, and the quadratic
+ * behaviour of using a naive repeated concatenation scheme.
+ */
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+typedef struct {
+    PyObject *large;  /* A list of previously accumulated large strings */
+    PyObject *small;  /* Pending small strings */
+} _PyAccu;
+
+PyAPI_FUNC(int) _PyAccu_Init(_PyAccu *acc);
+PyAPI_FUNC(int) _PyAccu_Accumulate(_PyAccu *acc, PyObject *unicode);
+PyAPI_FUNC(PyObject *) _PyAccu_FinishAsList(_PyAccu *acc);
+PyAPI_FUNC(PyObject *) _PyAccu_Finish(_PyAccu *acc);
+PyAPI_FUNC(void) _PyAccu_Destroy(_PyAccu *acc);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* Py_ACCU_H */
+#endif /* Py_LIMITED_API */