Misc perf fixes (toml, yaml)

[pjfanning]

suggested by Claude AI

Performance Fixes

File	Fix	Impact
yaml/YAMLGenerator.java:691	String.valueOf(v.toString()) → v.toString() — eliminated redundant String conversion on BigInteger writes	High
yaml/YAMLGenerator.java:327	Long.valueOf(id).toString() → Long.toString(id) — avoids unnecessary boxing	Low
yaml/YAMLGenerator.java:799	Eliminated Optional.of().map().Anchor() chain → direct new Anchor(id.toString())	Medium
toml/TomlParser.java:328-336	Replaced new String().replace().toCharArray() (3 allocations) with single-pass in-place char copy skipping underscores	High
toml/TomlParser.java:450	Added indexOf('_') guard before .replace("_", "") — avoids unnecessary String copy on the common case (no underscores)	Medium

[github-actions]

🧪 Code Coverage Report

Metric	Coverage	Change
Instructions	78.75%	📈 +0.010%
Branches	72.67%	📈 +0.030%

Coverage data generated from JaCoCo test results

[github-actions]

🧪 Code Coverage Report

Metric	Coverage	Change
Instructions	78.80%	📈 +0.010%
Branches	72.69%	📈 +0.030%

Coverage data generated from JaCoCo test results

[skrcode]

I ran JAIPilot Cloud against this exact PR head. It found one additional TOML allocation cleanup: compact underscores in place in the parser-owned token buffer instead of allocating a second char array.

The same 113 focused parser tests passed before and after, including a long token that forces buffer growth and a following token that checks for corruption. Five fixed allocation batches on a 10,000-digit underscored integer moved from 551,104 to 524,416 allocated bytes per parse, and the full 1,558-test clean build passed.

PR directly onto this source branch: pjfanning#13
Cloud-generated draft and full evidence: skrcode#2

Only the underscore slow path changes; ordinary integer tokens are untouched, and the long-token fixture is not presented as a production-throughput claim.

[github-actions]

🧪 Code Coverage Report

Metric	Coverage	Change
Instructions	78.80%	📈 +0.010%
Branches	72.69%	📈 +0.030%

Coverage data generated from JaCoCo test results

[pjfanning]

Thanks @skrcode - that change looks sensible and since it improves the change in my PR, it makes sense to merge it into this.