fix(json): parse deep nesting with a heap-backed stack

[proggeramlug]

Summary

Closes #7817. JSON.parse now switches from the recursive fast path to the existing flat tape representation once nesting passes the worker-stack threshold, then materializes that tape with an explicit heap-backed work stack. The reported 300,000-level document parses on a 2 MiB worker stack.

A separate 500,000-level iterative nesting budget keeps resource use finite and reports a catchable RangeError.

Changes

• make tape validation strict enough to own the deep-input path without the recursive serde validation pass
• add iterative tape materialization for arrays, objects, duplicate keys, strings, numbers, booleans, and null
• route all three parse entries through the same recursive/iterative decision
• retain the optimized recursive and typed-shape paths for ordinary payloads

Related issue

Closes #7817. Follows the crash guard in #7816.

Test plan

• baseline reproduction: the old 1,001-level test passes on main by asserting that parsing returns an error
• cargo test -p perry-runtime json::tests::nesting_depth -- --nocapture
• cargo test -p perry-runtime json::tests -- --nocapture
• cargo test -p perry-runtime json_tape::tests -- --nocapture
• full perry-runtime suite: 2,108 passed; one unrelated promise keyed-table timing assertion failed under concurrent local load and passed immediately when rerun alone
• cargo fmt --all -- --check
• git diff --check
• scripts/check_file_size.sh still reports only the two pre-existing main offenders fixed by fix(lint): unbreak the 2000-line file-size gate on main #7830

The quiet-mini benchmark is pending because its required lock is held by strcat-agent; no performance numbers are claimed here yet.

Screenshots / output

N/A (runtime parser change).

Checklist

• Added regression tests
• Added changelog.d/7832-iterative-json-parse.md
• Kept ordinary JSON on the existing optimized path
• No version bump or CHANGELOG.md edit
• Run ordinary-payload timing on the quiet mini when its lock is free

Summary by CodeRabbit

• New Features

  • JSON parsing now supports deeply nested documents, including inputs with up to 500,000 levels of nesting.
  • Deep JSON structures can be parsed reliably even with limited stack space.

• Bug Fixes

  • Improved validation rejects malformed numbers, invalid escapes, control characters, trailing content, and incomplete structures.
  • Excessively deep input now returns a clear RangeError instead of risking parser failure.