Use Unpooled in StreamedRequest#onNext

[dagnir]

Description

the onNext method can be called from a non SDK owned thread, and if we allocate
the ByteBuf using the pooling allocator, then we pollute the customer's thread
with a ThreadLocal.

Using Unpooled#wrappedBuffer avoids actually copying the arrays so this should
be more performant as well.

Motivation and Context

Fixes #1133

Testing

Screenshots (if appropriate)

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)

Checklist

• I have read the CONTRIBUTING document
• Local run of mvn install succeeds
• My code follows the code style of this project
• My change requires a change to the Javadoc documentation
• I have updated the Javadoc documentation accordingly
• I have read the README document
• I have added tests to cover my changes
• All new and existing tests passed
• A short description of the change has been added to the CHANGELOG
• My change is to implement 1.11 parity feature and I have updated LaunchChangelog

License

• I confirm that this pull request can be released under the Apache 2 license

[codecov-io]

Codecov Report

Merging #1136 into master will decrease coverage by 0.02%.
The diff coverage is 100%.

@@             Coverage Diff              @@
##             master    #1136      +/-   ##
============================================
- Coverage     58.77%   58.75%   -0.03%     
+ Complexity     4557     4556       -1     
============================================
  Files           742      742              
  Lines         22941    22895      -46     
  Branches       1713     1707       -6     
============================================
- Hits          13483    13451      -32     
+ Misses         8770     8754      -16     
- Partials        688      690       +2

Impacted Files	Coverage Δ	Complexity Δ
.../http/nio/netty/internal/NettyRequestExecutor.java	64.24% <100%> (-0.4%)	29 <0> (ø)
...nio/netty/internal/OldConnectionReaperHandler.java	81.81% <0%> (-9.1%)	12% <0%> (-1%)
...awssdk/codegen/poet/client/specs/ProtocolSpec.java	81.25% <0%> (-3.75%)	11% <0%> (ø)
...mazon/awssdk/core/internal/retry/RetryHandler.java	79.54% <0%> (-2.28%)	11% <0%> (ø)
.../amazon/awssdk/http/apache/ProxyConfiguration.java	40.38% <0%> (-1.13%)	16% <0%> (ø)
...wssdk/http/nio/netty/internal/ResponseHandler.java	62.58% <0%> (-0.48%)	15% <0%> (ø)
...awssdk/regions/internal/util/EC2MetadataUtils.java	22.9% <0%> (-0.43%)	8% <0%> (ø)
...mazon/awssdk/protocols/ion/internal/IonParser.java	76.42% <0%> (-0.29%)	67% <0%> (ø)
...re/amazon/awssdk/http/apache/ApacheHttpClient.java	63.97% <0%> (-0.23%)	17% <0%> (ø)
...ttp/pipeline/stages/MakeAsyncHttpRequestStage.java	80.21% <0%> (-0.22%)	14% <0%> (ø)
... and 22 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 9f5ae80...eb940bd. Read the comment docs.

[spfink]

Any testing we can add here?

[dagnir]

@spfink Can't think of a good one. This avoids creating the ThreadLocal(s) used by the memory pooling mechanism in Netty so ideally we would check that they aren't created but not sure how besides using reflection to to get the TreadLocal instance and calling get()...