stream: inline unbuffered _write

[ronag]

A ~15% improvement for the sync case

 streams/writable-manywrites.js callback='no' writev='no' sync='no' n=2000000 1.68 % ±5.45% ±7.25% ±9.45%
streams/writable-manywrites.js callback='no' writev='no' sync='yes' n=2000000 *** 15.98 % ±3.86% ±5.14% ±6.69%
streams/writable-manywrites.js callback='no' writev='yes' sync='no' n=2000000 0.28 % ±5.36% ±7.14% ±9.30%
streams/writable-manywrites.js callback='no' writev='yes' sync='yes' n=2000000 0.89 % ±4.58% ±6.09% ±7.92%
streams/writable-manywrites.js callback='yes' writev='no' sync='no' n=2000000 -1.19 % ±4.94% ±6.58% ±8.56%
streams/writable-manywrites.js callback='yes' writev='no' sync='yes' n=2000000 *** 19.24 % ±4.17% ±5.55% ±7.23%
streams/writable-manywrites.js callback='yes' writev='yes' sync='no' n=2000000 -1.97 % ±5.91% ±7.87% ±10.24%
streams/writable-manywrites.js callback='yes' writev='yes' sync='yes' n=2000000 -0.03 % ±4.68% ±6.22% ±8.10%

Checklist

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

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/30793/

[himself65]

LGTM,
and should we add a comment like // inline doWrite(stream, state, false, len, chunk, encoding, cb) for better understand

[ronag]

and should we add a comment like // inline doWrite(stream, state, false, len, chunk, encoding, cb) for better understand

It's not exactly the same anymore, since some of the code that is not relevant is removed. So it might make it more confusing?

[ronag]

Landed in fd10be4