stream: use state.ending to see if stream called end()

[MoonBall]

Calling writable.end() will probably synchronously call
writable.write(), in such a situation the state.ended
is false and writable.write() doesn't trigger writeAfterEnd().

There is an example as below:

const { Writable } = require('stream');

let data = '';
const w = new Writable({
  write: function(chunk, encoding, cb) {
    process.nextTick(() => {
      data += chunk;
      cb();
    });
  },
});

w.on('prefinish', () => {
  // w.write('prefinish');
});
w.on('finish', () => {
  w.write('finish');
});

w.end(() => {
  console.log(data);
});

when 'prefinish' listener doesn't call w.write(), w.end() will synchronously emit 'finish' event. The console's output is: finish.
when 'prefinish' listener calls w.write(), the 'finish' event will be asynchronously emitted. The w.write() that called by the 'finish' listener will trigger writeAfterEnd() and throw a Error.

I think using state.ended to decide whether to trigger writeAfterEnd is diffcult to understand and is somewhat unreasonable.

Checklist

• make -j4 test (UNIX), or vcbuild test (Windows) passes
• tests and/or benchmarks are included
• commit message follows commit guidelines

Affected core subsystem(s)

stream

[mcollina]

I do not understand what is the current behavior and what you want to achieve with this change. Can you please edit the description of this PR to clarify that? Examples works better than a textual form.

Can you please add/change the unit tests?

[MoonBall]

@mcollina I added test cases. are they right?

[mcollina]

CI: https://ci.nodejs.org/job/node-test-pull-request/12581/

[mafintosh]

This is how i would expect things to work today and I would consider this a bug

[MoonBall]

@mcollina There are some failed checks. What should I do?

[mcollina]

I think it's CI that is flaky.

CITGM: https://ci.nodejs.org/view/Node.js-citgm/job/citgm-smoker/1208/

[addaleax]

I would feel okay with considering this semver-major just to be careful (but also okay with not doing so).

[mcollina]

We can land it semver-minor and taking it some more time before backporting.

[mcollina]

CITGM: https://ci.nodejs.org/view/Node.js-citgm/job/citgm-smoker/1210/

[mcollina]

Can you please rebase on top of master?

[MoonBall]

Rebased.

[mcollina]

CITGM: https://ci.nodejs.org/view/Node.js-citgm/job/citgm-smoker/1216/

[mcollina]

CI: https://ci.nodejs.org/job/node-test-pull-request/12801/
CITGM (master): https://ci.nodejs.org/view/Node.js-citgm/job/citgm-smoker/1241/
CITGM: https://ci.nodejs.org/view/Node.js-citgm/job/citgm-smoker/1242/

[mcollina]

LGTM

[mcollina]

CITGM seems similar, landing.

[mcollina]

Landed as c7ca07a