perf: speed up constant string joins

[He-Pin]

Summary

• Keep large constant std.makeArray(... literal) arrays as an O(1) constant array view.
• Fast-path std.join over constant/repeated string arrays with exact StringBuilder sizing.
• Mark ASCII JSON-safe joined/repeated strings so byte rendering can skip escape scanning and UTF-8 encoding.
• Add JVM/Native UTF-16-lane SWAR ascii-safe checks; Scala.js keeps a scalar fallback.
• Drop remaining private[this] in touched Scala sources.

Benchmarks

Baseline: latest upstream/master at 8b67cb1e. This PR: 9e7c7f13.

JMH + GC profiler for bench/resources/cpp_suite/large_string_join.jsonnet, lower is better:

Metric	upstream/master	This PR	Delta
runtime	0.540 ms/op	0.271 ms/op	-49.8%
gc.alloc.rate.norm	1,531,272 B/op	632,976 B/op	-58.7%

Scala Native hyperfine against latest source-built jrsonnet master 5b43fa8 (jrsonnet 0.5.0-pre98), lower is better:

Command	Mean
sjsonnet native, upstream/master	5.3 +/- 1.6 ms
sjsonnet native, this PR	4.6 +/- 1.1 ms
latest jrsonnet	8.0 +/- 1.3 ms

The Native CLI run is short and outlier-heavy, but the current mean has this PR ahead of latest jrsonnet on large_string_join. The stronger signal is the JMH runtime/allocation reduction above.

Validation

• ./mill -i __.checkFormat
• git diff --check
• ./mill -i 'sjsonnet.jvm[3.3.7].test'
• ./mill -i 'sjsonnet.js[3.3.7].test'
• ./mill -i 'sjsonnet.js[3.3.7].compile' 'sjsonnet.native[3.3.7].compile'
• ./mill -i 'sjsonnet.native[3.3.7].nativeLink'
• ./mill -i bench.runJmh sjsonnet.bench.RegressionBenchmark.main -p path=bench/resources/cpp_suite/large_string_join.jsonnet -prof gc
• hyperfine --warmup 10 --min-runs 50

References

• Head: 9e7c7f1330230966f65dbdd4a5765015b6826366
• Latest source-built jrsonnet: 5b43fa88b8c43856dd5a2daa9c5c251153c5e14d

[He-Pin]

Closing as superseded by #828, which carries the TOML and repeated join work with current docs-aligned benchmark data.

[He-Pin]

Reopened and moved back to draft rather than closing. It is not negative, but it is currently superseded by #828, so it should not be ready for review unless we decide to keep it as a separate smaller join-only PR.