perf: bypass query plans for JSON reads

[chiinlquah]

What changes are proposed in this pull request?

Read JSON files in PlanBasedJsonHandler through the typed ReadBytes I/O operation and decode
the returned bytes directly with Arrow JSON. This avoids constructing and executing a generic
single-node query plan for JSON scans while preserving the generic query-plan path for other plan
shapes.

The Arrow JSON byte decoder is shared with SyncJsonHandler. A focused test wraps the plan
executor and rejects QueryPlan operations, proving that JSON reads take only the typed I/O path.

In a local CommitRange benchmark over 1,000 one-action commit files (50 warmups, 10 measured
iterations, release/O3 native build), the median fell from 3.616 s with generic driver-local plan
execution to 0.788 s with direct JSON, a 4.59x speedup. Both variants launched zero Spark jobs.
The remaining cost includes three scans per commit in the benchmarked CommitRange call path.

How was this change tested?

• cargo test -p delta_kernel --features declarative-plans engine::plans::json::tests::test_read_json_files
• Repository pre-commit checks (formatting, Clippy, and tests)

[codecov]

Codecov Report

❌ Patch coverage is 86.79245% with 7 lines in your changes missing coverage. Please review.
✅ Project coverage is 90.34%. Comparing base (048ffbe) to head (0adc981).

Files with missing lines	Patch %	Lines
kernel/src/engine/plans/json.rs	89.74%	1 Missing and 3 partials ⚠️
kernel/src/engine/arrow_utils/mod.rs	76.92%	0 Missing and 3 partials ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #3314   +/-   ##
=======================================
  Coverage   90.34%   90.34%           
=======================================
  Files         250      250           
  Lines       89131    89172   +41     
  Branches    89131    89172   +41     
=======================================
+ Hits        80524    80563   +39     
- Misses       5672     5673    +1     
- Partials     2935     2936    +1     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.