feat(java): parse quoted JSON scalar values

[chaokunyang]

Why?

Financial and other APIs commonly encode boolean and numeric values as JSON strings. Fory JSON
previously rejected those values for declared scalar targets, so users had to register custom
codecs even for standard types such as BigDecimal.

What does this PR do?

• Accept quoted ordinary token text for declared boolean, byte, short, int, long, float, double,
  BigInteger, and BigDecimal targets without a configuration switch.
• Parse quoted values through the same representation-specific token parsers as native JSON
  tokens, without allocating an intermediate String or another per-value carrier.
• Keep natural dynamic JSON typing: a quoted value read as Object remains a String; a quoted
  numeric value read as Number is a Double.
• Cover roots, generated and interpreted object codecs, arrays, collections, maps, Latin1, UTF-8,
  UTF-16, arbitrary-precision limits, and malformed quoted input.
• Document the user-visible coercion behavior.

Related issues

N/A.

AI Contribution Checklist

• Substantial AI assistance was used in this PR: yes / no
• If yes, I included a completed AI Contribution Checklist in this PR description and the required AI Usage Disclosure.
• If yes, my PR description includes the required ai_review summary and screenshot evidence or equivalent persisted links of the final clean AI review results from both fresh reviewers described in AI_POLICY.md, the Fory-guided reviewer and the independent general reviewer, on the current PR diff or current HEAD after the latest code changes.

Does this PR introduce any user-facing change?

Yes. Declared boolean and numeric JSON targets now accept their ordinary token text inside a JSON
string. Serialization output remains unchanged.

• Does this PR introduce any public API change?
• Does this PR introduce any binary protocol compatibility change?

Benchmark

Environment: macOS arm64, OpenJDK 25.0.3, JMH 1.37, one fork/thread, -Xms2g -Xmx2g, one-second
iterations, and -prof gc.

Unquoted generated primitive POJO parsing against the exact merge-base artifact:

Input	Baseline	Current	Change	Allocation
Latin1	60.159 ns/op	58.797 ns/op	-2.263%	40.000 / 40.000 B/op
UTF-8	56.840 ns/op	55.550 ns/op	-2.270%	40.000 / 40.000 B/op
UTF-16	131.191 ns/op	133.706 ns/op	+1.917%	40.001 / 40.001 B/op

Quoted and unquoted normalized allocations are equal to JMH counter precision:

Case	Unquoted	Quoted
BigDecimal	184.001 B/op	184.001 B/op
BigInteger	264.001 B/op	264.002 B/op
Combined scalar POJO	592.003 B/op	592.003 B/op

Final-head targeted validation:

ENABLE_FORY_DEBUG_OUTPUT=1 mvn -q -pl fory-json -Dtest=JsonScalarTest \
  spotless:check checkstyle:check test

139 tests passed with no failures, errors, or skips.