* And pycore_lifecycle.h and pycore_pathconfig.h headers to
  Include/internal/
* Move Py_BUILD_CORE specific code from coreconfig.h and
  pylifecycle.h to pycore_pathconfig.h and pycore_lifecycle.h
* Move _Py_wstrlist_XXX() definitions and _PyPathConfig code
  from pycore_state.h to pycore_pathconfig.h
* Move "Init" and "Fini" function definitions from pylifecycle.c to
  pycore_lifecycle.h.

The accu.h header is no longer part of the Python C API: it has been
moved to the "internal" headers which are restricted to Python
itself.

Replace #include "accu.h" with #include "pycore_accu.h".

If Py_BUILD_CORE is defined, the PyThreadState_GET() macro access
_PyRuntime which comes from the internal pycore_state.h header.
Public headers must not require internal headers.

Move PyThreadState_GET() and _PyInterpreterState_GET_UNSAFE() from
Include/pystate.h to Include/internal/pycore_state.h, and rename
PyThreadState_GET() to _PyThreadState_GET() there.

The PyThreadState_GET() macro of pystate.h is now redefined when
pycore_state.h is included, to use the fast _PyThreadState_GET().

Changes:

* Add _PyThreadState_GET() macro
* Replace "PyThreadState_GET()->interp" with
  _PyInterpreterState_GET_UNSAFE()
* Replace PyThreadState_GET() with _PyThreadState_GET() in internal C
  files (compiled with Py_BUILD_CORE defined), but keep
  PyThreadState_GET() in the public header files.
* _testcapimodule.c: replace PyThreadState_GET() with
  PyThreadState_Get(); the module is not compiled with Py_BUILD_CORE
  defined.
* pycore_state.h now requires Py_BUILD_CORE to be defined.

* Rename Include/internal/ header files:

  * pyatomic.h -> pycore_atomic.h
  * ceval.h -> pycore_ceval.h
  * condvar.h -> pycore_condvar.h
  * context.h -> pycore_context.h
  * pygetopt.h -> pycore_getopt.h
  * gil.h -> pycore_gil.h
  * hamt.h -> pycore_hamt.h
  * hash.h -> pycore_hash.h
  * mem.h -> pycore_mem.h
  * pystate.h -> pycore_state.h
  * warnings.h -> pycore_warnings.h

* PCbuild project, Makefile.pre.in, Modules/Setup: add the
  Include/internal/ directory to the search paths of header files.
* Update includes. For example, replace #include "internal/mem.h"
  with #include "pycore_mem.h".

Don't call _Py_FatalError_PrintExc() nor flush_std_files() if the
current thread doesn't hold the GIL, or if the current thread
has no Python state thread.

* Add #include "internal/mem.h" to C files using
  _PyMem_SetDefaultAllocator().
* Include/internal/mem.h now requires Py_BUILD_CORE to be defined.

_io.IncrementalNewlineDecoder's initializer possibly assigns out-of-range
value to the bitwise struct field.

The root widget was accessed as a global variable in the Application's method.

Add pyatomic.h to the VS project (it wasn't referenced).

* Remove _PyThreadState_Current
* Replace GET_TSTATE() with PyThreadState_GET()
* Replace GET_INTERP_STATE() with _PyInterpreterState_GET_UNSAFE()
* Replace direct access to _PyThreadState_Current with
  PyThreadState_GET()
* Replace _PyThreadState_Current with
  _PyRuntime.gilstate.tstate_current
* Rename SET_TSTATE() to _PyThreadState_SET(), name more
  consistent with _PyThreadState_GET()
* Update outdated comments

Convert _Py_Dealloc() macro into a static inline function. Moreover,
it is now also defined as a static inline function if Py_TRACE_REFS
is defined.

Make _PySys_AddXOptionWithError() and _PySys_AddWarnOptionWithError()
functions private again. They are no longer needed to initialize Python:
_PySys_EndInit() is now responsible to add these options instead.

Moreover, PySys_AddWarnOptionUnicode() now clears the exception on
failure if possible.

* bpo-34523, bpo-34403: Fix config_init_fs_encoding(): it now uses
  ASCII if _Py_GetForceASCII() is true.
* Fix a regression of commit b2457efc.
* Fix also a memory leak: get_locale_encoding() already allocates
  memory, no need to duplicate the string.

Use curses.ncurses_version for conditionally skipping a test.

It was overridden by the lineno of the first decorator. Now it is
the lineno of 'def' or 'class'.

Needs backport to 3.7. In 3.6 the description is correct.

Some FreeBSD buildbots fail to run this test as the eof was not being received by the server if the size is not big enough. This behaviour only appears if the client is using TLS1.3.

Fix 'install-sh' file permission: add execution bit ("chmod +x"),
"-rw-rw-r--." becomes "-rwxrwxr-x.".

After commit d0f49d2f, the output of the
test suite is always buffered as the test output needs to be included in
the JUnit file in same cases (as when a test fails). This has the
consequence that printing or using debuggers (like pdb) in the test
suite does not result in a good user experience anymore.

This commit modifies the test suite runner so it only captures the test
output when the JUnit file is requested to fix the regression so prints
and debuggers are usable again.

Convert Py_XINCREF() and Py_XDECREF() macros into static inline
functions.

* Include memo in the documented signature of copy.deepcopy()

The memo argument is mentioned lower on the doc page under writing a
`__deepcopy__` method, but is not included in the documented function signature.
This makes it easy to miss, and can lead to incorrect/buggy implementations of
`__deepcopy__` -- which is exatly what just happpend to me!

Modules imported last are now cleared first at interpreter shutdown.

A newly imported module is moved to the end of sys.modules, behind
modules on which it depends.

"static inline" should be used directly. Forcing the compiler to
inline is not recommended.

* Convert Py_INCREF() and Py_DECREF() macros into static inline
  functions.
* Remove _Py_CHECK_REFCNT() macro: code moved into Py_DECREF().

Topics include opening, title and status bar, .py* extension, and running.

In commit 372d705d a new optimization to the list() type was introduced but it was not added
to the optimization section in What's new for Python 3.8.

unittest.mock now includes mock calls in exception messages if
assert_not_called, assert_called_once, or assert_called_once_with
fails.

Add a paragraph in "Help and preferences", "Help sources" subsection.

The list() constructor isn't taking full advantage of known input
lengths or length hints. This commit makes the constructor
pre-size and not over-allocate when the input size is known (the
input collection implements __len__). One on the main advantages is
that this provides 12% difference in memory savings due to the difference
between overallocating and allocating exactly the input size.

For efficiency purposes and to avoid a performance regression for small
generators and collections, the size of the input object is calculated using
__len__ and not __length_hint__, as the later is considerably slower.