[WIP] ByteString Two-Way string matching in indexOfSlice

[pjfanning]

• See [WIP] use getLong when matching byte sequences in ByteString #2323 and Revisit indexOfSlice performance improvements #2321
• the aim is to benchmark both solutions - this one is more complicated so it would need to be much better to justify being used

[Copilot]

Pull request overview

This PR introduces a Netty-derived two-way string matching algorithm for ByteString.indexOfSlice (and a shared byte-segment matcher), aiming to evaluate potential performance improvements versus existing implementations.

Changes:

• Add Netty-derived two-way substring search implementation to ByteString1 / ByteString1C indexOfSlice.
• Introduce optimized byte-segment matching helpers (bytesMatch / arrayBytesMatch) using getLong comparisons.
• Update licensing files to reflect Netty-derived code usage in ByteString.scala.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 4 comments.

File	Description
LICENSE	Adds ByteString.scala to the list of Netty-derived sources.
legal/pekko-actor-jar-license.txt	Mirrors the LICENSE update for the actor JAR legal file.
actor/src/main/scala/org/apache/pekko/util/SWARUtil.scala	Makes SWAR helpers final and adds Netty-derived helpers (arrayBytesMatch, maxSuf).
actor/src/main/scala/org/apache/pekko/util/ByteString.scala	Adds Netty-derived two-way matching for indexOfSlice and introduces bytesMatch abstraction/optimizations.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[He-Pin]

[WIP] The Two-Way string matching approach from Netty is a good candidate for optimizing indexOfSlice on large ByteStrings. However:

1. This is marked WIP and the PR description says 'the aim is to benchmark both solutions'. Without benchmark results showing a clear win, this is hard to evaluate.
2. The licensing update (adding Netty-derived code to LICENSE) is correctly handled.
3. This relates to #2323 which takes a simpler approach (just using getLong for byte matching). Consider benchmarking both approaches and merging whichever wins.

Recommendation: Complete the benchmarks and either merge or close based on results. Don't leave as WIP indefinitely.