object is destroyed. The destructor now closes the file if needed. The close()
method can now be called twice: the second call does nothing.

- Use support.captured_stderr() where appropriate
- Removes some "from test.support import xxx" import and uses support.xxx
  instead.

Handle objects are created. Pass the traceback to call_exception_handler() in
the 'source_traceback' key.

The traceback is truncated to hide internal calls in asyncio, show only the
traceback from user code.

Add tests for the new source_traceback, and a test for the 'Future/Task
exception was never retrieved' log.

object was created to the "coroutine ... was never yield from" log

This makes the behavior match that of Parser.  Patch by Vajrasky Kok.

move two functions next to the functions that use them.

- Python issue 21163: Fix more "Task was destroyed but it is pending!" logs in
  tests
- Add test to check that run_until_complete() checks the loop of the future

- Uniformize repr() output to format "<Class ...>"
- On Python 3.5+, repr(Task) uses the qualified name instead of the short name
  of the coroutine

files closing.  Patch by Claudiu Popa.

Specifically, it was relying a URL which gave HTTP 403 and used it to assert
it's methods, this changes undoes that and provides a local http server with
similar properties.

Patch contributed by Vajrasky Kok.

parameters on encodestring and decodestring.

Patch by Petri Lehtinen.

Issue #19145:  Fix handling of negative values for a "times" keyword argument to itertools.repeat()>

(Patch contributed by Vajrasky Kok.)

fit in the default width of the text window (45 chars).

done: use the first line number of the code object instead of the current line
number of the generator frame.

The name of the coroutine is not enough because many coroutines may have the
same name. It's a common case in asyncio tests for example.