stream: avoid per-chunk promises in webstream adapters

[mcollina]

Continuing the WHATWG streams optimization work, this round targets the stream/webstream adapters (Readable.fromWeb(), Writable.fromWeb(), Duplex.fromWeb()), which had never been profiled.

Readable.fromWeb() and the read side of Duplex.fromWeb() allocated a promise, a read-result object, and two reaction closures for every chunk through reader.read(). Only one read is ever in flight (_read is not called again before push()), so a single reused read request now delivers chunks through readableStreamDefaultReaderRead(), forwarding each chunk in a microtask to keep the previous delivery order relative to errors and destroy.

Writable.fromWeb() and the write side of Duplex.fromWeb() paid two derived promises off writer.ready plus the writer.write() promise and a fresh closure pair per chunk. A single shared write request (the same contract pipeTo uses since #64890) now dispatches chunks directly and settles the node callback. Failures are delivered in a microtask because the callback can destroy the stream while the writable machinery is mid-transition.

Also adds benchmark/webstreams/adapters.js — the suite had no rows for the adapter paths.

Results (30 runs):

                                                         confidence improvement accuracy (*)   (**)   (***)
webstreams/adapters.js kind='readable-from-web' n=100000          *      5.07 %       ±4.84% ±6.44%  ±8.39%
webstreams/adapters.js kind='readable-to-web' n=100000                   0.92 %       ±6.25% ±8.32% ±10.83%
webstreams/adapters.js kind='writable-from-web' n=100000        ***     39.27 %       ±7.19% ±9.58% ±12.48%
webstreams/adapters.js kind='writable-to-web' n=100000                  -1.74 %       ±6.71% ±8.94% ±11.63%

The two toWeb rows are untouched paths and neutral, included for coverage. Beyond the test suite and WPT, the change was validated with a differential stress harness (error mid-write, writev with cork, erroring controller, destroy during data, pre-closed/pre-errored streams, duplex echo, slow-sink backpressure): the observable event logs are byte-identical to the previous implementation.

[nodejs-github-bot]

Review requested:

• @nodejs/performance

[codecov]

Codecov Report

❌ Patch coverage is 90.62500% with 15 lines in your changes missing coverage. Please review.
✅ Project coverage is 90.12%. Comparing base (7aaf9b4) to head (5095bc9).

Files with missing lines	Patch %	Lines
lib/internal/webstreams/adapters.js	90.44%	14 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #65548      +/-   ##
==========================================
- Coverage   90.13%   90.12%   -0.01%     
==========================================
  Files         751      751              
  Lines      253639   253714      +75     
  Branches    47790    47797       +7     
==========================================
+ Hits       228618   228663      +45     
- Misses      16264    16309      +45     
+ Partials     8757     8742      -15     

Files with missing lines	Coverage Δ
lib/internal/webstreams/readablestream.js	87.21% <100.00%> (+<0.01%)	⬆️
lib/internal/webstreams/adapters.js	88.03% <90.44%> (+1.42%)	⬆️

... and 39 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.