stream: keep consumer state in fast mode

[mcollina]

Create null-prototype share and broadcast consumer state with fast
properties instead of V8 dictionary properties.

Assisted-by: Pi

[nodejs-github-bot]

Review requested:

• @nodejs/performance
• @nodejs/streams

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 90.30%. Comparing base (e7d8ab5) to head (350436a).
⚠️ Report is 12 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main   #66266   +/-   ##
=======================================
  Coverage   90.30%   90.30%           
=======================================
  Files         789      789           
  Lines      272880   272879    -1     
  Branches    52110    52109    -1     
=======================================
  Hits       246418   246418           
- Misses      16912    16921    +9     
+ Partials     9550     9540   -10     

Files with missing lines	Coverage Δ
lib/internal/streams/iter/broadcast.js	90.50% <100.00%> (ø)
lib/internal/streams/iter/share.js	89.30% <100.00%> (-0.02%)	⬇️

... and 19 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.

[mcollina]

Benchmarks:

Benchmark	Consumers	Before	After	Change
streams/iter-throughput-share	2	121,268	123,878	+2.2%
streams/iter-throughput-share	8	193,168	200,535	+3.8%
streams/iter-throughput-share	32	213,996	217,894	+1.8%
streams/iter-throughput-share-sync	2	776,705	839,246	+8.1%
streams/iter-throughput-share-sync	8	1,773,686	1,831,000	+3.2%
streams/iter-throughput-share-sync	32	2,497,892	2,923,947	+17.1%
streams/iter-throughput-broadcast	2	10,637	10,832	+1.8%
streams/iter-throughput-broadcast	4	8,415	8,654	+2.8%

Share used batches=10000 n=5 backpressure=unbounded; sync share used batches=10000 n=5; broadcast used api=iter datasize=1048576 n=500. The broadcast results are noisy, so I would not treat those small differences as conclusive.