fix[notask]: drain in-flight blob blocks before cleanup

[yuranich]

What problem does this PR solve?

Cancelled registry downloads can still persist Hypercore blocks after the range request is destroyed. Responses already in flight may arrive between destroy() and clear(), so the cache continues growing across distinct cancellations.

How does it solve it?

• Waits for Hypercore peer requests and data processing to drain before clearing and compacting the cancelled range, with a bounded timeout.
• On drain timeout, logs remaining per-peer inflight / dataProcessing counters and still clears (best-effort).
• Adds a real Corestore/Hyperblobs regression test that cancels while blocks are in flight, reopens storage, and verifies that zero blob blocks remain.
• Adds a real-storage timeout path that holds core.verify so dataProcessing stays nonzero past the drain deadline, plus a focused unit coverage of the timeout log payload.

Breaking changes

None.

Replaces #3496 (same change, opened from an upstream branch instead of a fork).

[github-actions]

Review Status

Current Status: ✅ APPROVED
Approvals so far: Team Lead: 1, Member: 1

[github-actions]

License compliance — clean

No new dependency license findings in this PR.

Warn-only (shadow) mode — this check does not block merges yet.

Updated automatically by the canonical license compliance workflow.

NOTICE presence (advisory)

Missing NOTICE (advisory, does not block):

• ./.github/actions/release-merge-guard
• ./packages/vla-ggml/sim/server
• ./packages/asr-ggml/benchmarks/server
• ./packages/embed-llamacpp/benchmarks/performance
• ./packages/embed-llamacpp/benchmarks/server
• ./packages/llm-llamacpp/benchmarks/performance
• ./packages/llm-llamacpp/benchmarks/server
• ./packages/ggml-coload-smoke
• ./packages/sdk/e2e
• ./packages/fabric/test/integration
• ./docs/website