fix(query): honor counter-function semantics in ExactAgg dispatch (closes #301, #300)

[zzylol]

Summary

The asap engine collapsed sum / sum_over_time / increase / rate — four semantically-distinct PromQL counter idioms — onto the same two ExactAgg(Sum) reducer paths, so all instant counter sums returned the same wrong number and rate was systematically under-reported (issue #301; subsumes #300). Post-#299 the agent streams per-window deltas, so the inputs are correct; only the query-side dispatch lost the function distinction.

The 4 layered bugs (all fixed)

1. Engine dispatch — only rate vs everything-else branched; sum / sum_over_time / increase / instant-sum all hit one path.
2. Lookback — instant sum used a 5-min default == the [5m] range, so on a <5min producer every idiom captured the same horizon → identical values.
3. Instant projection (biggest) — the instant branch took samples.last() (the most-recent window's delta) instead of the cumulative-since-storage value PromQL sum(counter) requires.
4. Rate divisor — evaluate_exact_agg_rate divided by the nominal range (300 for [5m]) regardless of actual data coverage, halving the rate when the producer ran < range.

Fix shape

• control_plane: extend OuterFn from {Plain, Rate} to the full counter-fn taxonomy {Plain, Rate, Increase, SumOverTime}, populated by the analyzer's trace_from_promql walker (inner-counter-idiom wins for composed sum by (..) (rate(..)); precedence via new set_counter_fn). Carried on ASAPTierCandidate.outer_fn.
• engine dispatch (instant + range surfaces): branch on the typed counter-fn — Rate → rate reducer; Increase → accumulate windows over the [t-r,t] clip into one cumulative number; Plain instant sum → accumulate ALL windows over the full storage horizon (t0=0) → cumulative-since-start; SumOverTime over a counter → capability-miss → archive (asap stores deltas and cannot reconstruct Σ-of-cumulative-samples — issue asap engine: ExactAgg(Sum) dispatch collapses sum/sum_over_time/increase/rate to same code path (function semantics lost) #301 decision (a)).
• reducer: evaluate_exact_agg gains an accumulate_windows flag (collapse per-window deltas to one cumulative sample for instant-sum/increase; matrix surface keeps it false). evaluate_exact_agg_rate divides by min(range_seconds, actual_coverage_span_seconds) via new SketchStore::exact_agg_coverage_bounds.

Multinode validation (4 zones, 10000 series @ 100 Hz; asap node2:9091 vs VictoriaMetrics baseline node2:8428), steady state

Query	asap	baseline	rel-err	was (#301)
sum by (zone) (rate(...[5m]))	997,413	999,987	0.26%	64%
topk(5, sum by (zone) (rate(...)))	997,413	tracks rate	—	—
sum by (zone) (sum_over_time(...[5m]))	capability-miss → archive	16.99B	refuses (decision a)	returned wrong 60M

The four idioms now return distinct values (sum ≠ increase ≠ rate), confirming the dispatch no longer collapses them. Instant sum(counter) is cumulative-since-storage (runtime-dependent; residual gap is producer-runtime / flush-lag, not a dispatch bug). Working queries quantile_over_time(0.99, ...) (n=10000) and max by (zone) (quantile_over_time(...)) (#297) unaffected.

Test plan

• cargo build -p control_plane -p data_plane clean
• cargo test -p control_plane --lib → 783 passed
• cargo test -p data_plane --lib → 755 passed
• New unit/integration coverage: instant-sum accumulates all windows (not last); increase accumulates without divisor; sum_over_time capability-misses; rate divisor uses actual coverage
• Multinode rate rel-err 64% → 0.26%
• (pre-existing, unrelated) controller_plan_to_query_full_roundtrip_hll fails on origin/main too — not touched by this PR

🤖 Generated with Claude Code