https://bugs.python.org/issue35782

TextTestRunner of unittest.runner now uses time.perf_counter() rather
than time.time() to measure the execution time of a test: time.time()
can go backwards, whereas time.perf_counter() is monotonic.

Similar change made in libregrtest, pprint and random.

Currently, the *n* and *total* variables get converted to floats each time they are multiplied by random().  This minor tweak does the conversion just once and gets a small speedup (approx 3%).

Handle subnormal weights for choices()

random() takes precedence over getrandbits() if defined later
in the class tree.

bpo-31482: Missing bytes support for random.seed() version 1 #3614

Instead use keyword only arguments to os.register_at_fork for each of the scenarios.
Updates the documentation for clarity.

* bpo-16500: Allow registering at-fork handlers

* Address Serhiy's comments

* Add doc for new C API

* Add doc for new Python-facing function

* Add NEWS entry + doc nit

Issue 28475:  Improve error message for random.sample() with k < 0. (Contributed by Francisco Couzo).

Issue #27776: The os.urandom() function does now block on Linux 3.17 and newer
until the system urandom entropy pool is initialized to increase the security.

This change is part of the PEP 524.

to fully cover its state space.

- my previous removal and hg resolve mark had still left them and hooks did not catch it too!

Addresses issue #14927

error messages and comments.