feat(miner): consume ORB's live gate thresholds in self-review-context.js when available

[JSONbored]

Context

#6209 decided the design (see its pinned decision comment). The ORB-side read endpoint (sibling issue) exposes live gate thresholds; this issue is the AMS-side consumer in packages/loopover-miner/lib/self-review-context.js that calls it.

Requirements

• Before falling back to the existing static-.loopover.yml reconstruction, probe the target repo's ORB instance via the new endpoint (sibling issue). Reuse resolveGitHubToken's existing auth posture (packages/loopover-miner/lib/github-token-resolution.js) for this call rather than inventing new credential handling — this is additive to the same request flow AMS already makes against loopover's backend.
• Purely additive, fallback-preserving: if the probe 403s (not allowlisted), times out, or the target repo has no live override, fall back to today's static-file reconstruction with zero behavior change. AMS's fully-standalone (ORB-absent) path must not change at all.
• Short timeout on the probe (a few hundred ms) — this must never make discover/gate-prediction meaningfully slower for a repo where ORB isn't installed or doesn't respond.
• When the live thresholds ARE available, prefer them over the static reconstruction for the confidence_floor/scope_cap_files/scope_cap_lines fields specifically — leave every other predicted-gate field sourced from the static file exactly as today.

Deliverables

• The probe + fallback logic in self-review-context.js.
• Unit tests: ORB endpoint returns live thresholds → used in the prediction; endpoint 403s/times out/repo has no override → falls back to static reconstruction unchanged; fully-standalone repo (no .loopover.yml at all, no ORB) → unaffected, matches today's behavior exactly.

Test Coverage Requirements

This touches packages/loopover-miner/** — confirm current coverage convention for this package (check its own package.json/CI config for whether Codecov's 99%+ patch gate applies here the same way as src/**, or whether it's measured separately) and match it; do not skip tests on the assumption coverage doesn't apply.

Expected Outcome

AMS's gate-prediction becomes more accurate against a repo where ORB is actually installed and live-tuned, with zero behavior change for every other case (ORB absent, endpoint unreachable, repo not allowlisted).

Links & Resources

• Spec: upgrade AMS's predicted-gate to query ORB's live self-tuned config when both are installed #6209 (design decision — read its pinned comment before starting)
• Epic: AMS + ORB combined-use synergy features (both installed on the same repo) #6206 (parent epic)
• packages/loopover-miner/lib/self-review-context.js (today's static-file fallback, to be preserved)
• packages/loopover-miner/lib/github-token-resolution.js (resolveGitHubToken, the auth posture to reuse)
• The ORB-side endpoint sibling issue (produces the data this consumes)

[linear-code]

JSO-4451