feat(query-engine): computed top-k in range queries + stage-E equivalence tests (#581 stage E prep)

[milindsrivastava1997]

Summary

• Range queries previously had zero top-k support at all (execute_range_query_pipeline had no limiting/formatting params). Adds step-major top-k ranking/truncation (apply_range_topk) — ranks and truncates independently at each output timestamp, since a range query's surviving key set can differ step to step, unlike instant's single-value-per-group case. Wired through all three PromQL range call sites.
• Adds an instant/range equivalence test matrix ahead of stage E's full pipeline collapse: 4 window shapes (Tumbling ratio 1/2/3, one genuinely-overlapping Sliding case) × {no keys, SetAggregator keys, DeltaSetAggregator keys} × {instant, range}, plus 3 dedicated tests for the new topk plumbing (step-major re-ranking across steps, single-step-vs-instant equivalence, non-interference with fix(query-engine): range queries expand keys_query per output step #595's self-keyed sketch expansion).
• Two corrections to the originally-scoped test matrix, discovered while writing it (documented in the new test file's header): a Sliding window's lookback_ms must equal window_size_ms (active assert! in execute_range_query_pipeline, tracked separately as Sliding window execution: support multi-window merge/subtract in query engine #554), so "Sliding at ratio 2/3" isn't constructible; and SetAggregator/DeltaSetAggregator are keys-side types whose window_type is independent of the paired value aggregation's, so no cells needed excluding.
• Rebased onto main post-feat(store): added store interface for batched exact-window queries #627/refactor(query-engine): remove is_exact_query as a threaded bool #628 merge; no functional overlap (different regions of mod.rs/promql.rs).

Note: "computed top-k" over an arbitrary non-self-keyed expression (e.g. topk(5, rate(foo[5m]))) turns out to be unsupported everywhere in this codebase today, not just in range (SumAccumulator::query hard-errors on Statistic::Topk). Out of scope here, left as-is.

Test plan

• cargo test -p query_engine_rust --lib: 520/520 passing
• cargo clippy -p query_engine_rust --lib --tests -- -D warnings: clean
• Roborev review (job Re-architect code to use Redis for control messages #23): 1 Low finding fixed (dead-code retain), 1 Low finding (tie-break nondeterminism) left as-is per reviewer's own note — pre-existing instant-side behavior, not introduced here

🤖 Generated with Claude Code

[milindsrivastava1997]

Automated review findings.