[Feature #19630]

This dangerous behavior was removed in 4.0 (996cae65f3cc8fed60c6bb758b00882cac49389d)
but the documentation wasn't updated.

This appears to be a workaround for Windows XP behavior. It is unnecessary now.

Previously, calling IO#flush or closing an IO with unflushed buffered
writes would just invoke `#blocking_operation_wait` and flush the write
buffer using a `write` syscall. This change adds flushing through the
fiber scheduler by invoking the `#io_write` hook.

* Prefer IO::Buffer#write in IOScheduler

* Use Dir.tmpdir for test file

* Correctly handle errors in io_flush_buffer_fiber_scheduler

The changes are to `io.c` and `thread.c`.
I changed the API of 2 exported thread functions from `internal/thread.h` that
didn't look like they had any use in C extensions:

* rb_thread_wait_for_single_fd
* rb_thread_io_wait

I didn't change the following exported internal function because it's
used in C extensions:

* rb_thread_fd_select

I added a comment to note that this function, although internal, is used
in C extensions.

Fix io_pwrite fiber scheduler hook.

Re-organize page docs

[Bug #21703]

RUBY_CRASH_REPORT does not work in some cases when shelling out on Linux.
For example, given the following shell script dump.sh:

    #!/usr/bin/env bash
    cat > /tmp/crash

And we see it fails like this:

    $ RUBY_CRASH_REPORT="|dump.sh" ruby -rfiddle -e "Fiddle::Pointer.new(1, 10)[0]"
    cat: -: Resource temporarily unavailable