[GrantFox][Medium] Bound archive queries and historical reads

Summary

Bound archive queries and historical reads

Why this matters

Unbounded historical scans can exceed Soroban CPU and memory budgets.

Scope

Add cursor/page bounds, stable ordering, and gas-focused query tests.

Acceptance criteria

• Every archive query has a maximum page size.
• Cursors are deterministic and complete.
• Invalid cursors fail safely.
• Gas tests enforce the intended upper bound.

Validation

Add regression coverage for existing behavior, failure modes, authorization boundaries, and compatibility. The implementation must pass the repository CI checks.

Non-goals

• Typo-only, formatting-only, or documentation-only changes.
• Unrelated refactors or dependency upgrades.
• Weakening existing security, authorization, CI, or production safeguards.

Contributor application

Before implementation, comment with relevant experience, a 1–4 bullet approach, and an estimate for opening the first draft PR. Wait for maintainer assignment before coding.

PR requirements

Use a feature branch, include Closes #<issue-number>, check every acceptance criterion, link criteria to code/tests, explain security and failure-mode considerations, and pass CI.

Reward-readiness

This is a substantive GrantFox campaign issue. Merge and CI success do not by themselves guarantee reward eligibility; final reward-readiness is determined by campaign review.

First locate every archive query and historical-read entry point, then inspect existing behavior, failure-mode, authorization, and compatibility tests. Add regression and gas-focused coverage for bounded pages, deterministic complete cursors, and safely rejected invalid cursors. The work is done when all archive queries enforce maximum page sizes and the repository CI checks pass.