Several people have requested that the assert macros in EUnit should be
moved out to a separate header file. This patch puts them in
stdlib/include/assert.hrl, which gets included by the eunit.hrl file.
Thus, nothing changes for eunit users, but the asserts can now also be
included separately.

* dgud/stdlib/rand/OTP-12586:
  stdlib: Document and add normal distributed random value function
  stdlib: Add new random functionality/module

The old random module contains an old algorithm which have flaws
and the api requires the user to invoke seed and or checking if seed have
been invoked, if a non constant seed is to be used.

The api contains the following features:
- The user can invoke rand:unform/[0|1] directly and get a non constant seeding.
- The api is split in functional and non functional functions,
  i.e. usage of _s functions will not affect the process dictionary.
- The api contains several algorithms with different characteristics and
  can be extended with new algorithms in the future.
- Contains state of the art random number generators.
- Default algorithm is taylor made for erlang to be fast on 64bits machines.

Add two new system callbacks Module:system_get_state/1 and
Module:system_replace_state/2 to allow sys:get_state/1,2 and
sys:replace_state/2,3 to operate correctly even if a process is sys
suspended. Modify gen_server, gen_fsm, and gen_event to support the new
callbacks.

If a callback module does not export these functions, then by default the
Misc value (the same as that passed as the final argument to
sys:handle_system_msg/6, and returned as part of the return value of
sys:get_status/1,2) is treated as the callback module's state.

The previous behaviour of intercepting the system message and passing a
tuple of size 2 as the last argument to sys:handle_system_msg/6 is no
longer supported.

Add tests to verify the correctness of sys:get_state/1,2 and
sys:replace_state/2,3 when processes are sys suspended. Add two tests for
modules that implement special processes, one that exports
system_get_state/1 and system_replace_state/2 and one that doesn't.

Much of the credit for this patch goes to James Fish, who reported the
initial problem and implemented much of the fix.

In rest_for_one and one_for_all supervisors one child dying can cause
multiple children to be restarted. Previously if the child that caused
the restart is started successfully but another child fails to start,
the supervisor would not terminate this child with the other
successfully restarted children as no record of the pid was kept. Thus
the supervisor would try to start this child again. This could lead to
multiples of the same child or if the child is registered cause repeated
attempts at starting this child - until the max restart threshold was
reached.

Now the child that failed to start becomes the restarting child, instead
of staying with the same child, for the next restart attempt. This has
the following side effects:

1) In one_for_all the new version of the child that original died is
   terminated before a restart attempt is made.

2) In rest_for_one all succesfully restarted children are not terminated
   and restarting continues from the child that failed to start.

OTP-10106
OTP-10107