Indexer PR Feedback Allocation Follow-up

Status: Accepted
Date: 2026-02-02
Context:
- Review feedback highlighted unbounded allocations in indexer handlers and asked for clearer, live-memory guardrails.
- Allocation safety needed to remain cross-platform and avoid hard-coded assumptions.
- Test helper naming and error diagnostics in tests required clarification.
Decision:
- Add explicit allocation safety checks for request-driven string and vector allocations using the shared live-memory guard.
- Introduce a minimum-available-memory threshold and a cached-system entry point to avoid repeated probing where reuse is possible.
- Rename shared indexer test state helper and tighten ProblemDetails parsing in tests.
Consequences:
- Safer handling of request-sized allocations with clearer memory-policy documentation.
- Improved test helper clarity and more actionable test failures.
- Slightly more allocation checks per request, offset by the option to reuse a system snapshot.
Follow-up:
- Confirm code scanning alerts clear after the next GitHub Advanced Security scan.
- No migrations required.

Motivation

Close PR feedback on allocation safety and test clarity while keeping indexer handler behavior intact and aligned with live-memory guardrails.

Allocation sizing now checks request-derived bytes against live available memory before materializing strings or vectors.
The allocation guard exposes a cached-system entry point and enforces a minimum available memory threshold before allowing allocations.
Shared indexer test helpers use clearer naming and explicit expectations for response decoding.

Low risk: behavior is additive and defensive. Roll back by reverting this change set if allocation checks prove too strict in practice.