Add custom statistics (per-channel & cross-channel, multi-output) to StatsAggregator

[tombonfert]

Summary

Extends StatsAggregator in both the query engine and the reporting layer so users can inject arbitrary statistics functions alongside the built-ins (min/max/mean/median/start/end).

Two kinds of custom statistics

• Per-channel: computed once per input channel per interval, exactly like a built-in. Results ride the existing numeric_values map under the real channel names.
• Cross-channel: computed once per interval over a declared set of channels. Results carry a descriptor-provided channel_name (defaulting to the output label).

Self-describing descriptors

• PerChannelStatistic and CrossChannelStatistic (in the new custom_statistic.py) declare the function, its params, and mandatory aggregation_labels.
• A function returns a sequence of scalars mapped positionally to those labels, so one call can emit multiple outputs.
• Statistics are passed as lists of descriptors. Labels are the sole identity and must be globally unique and must not shadow built-ins.

Reporting integration

• custom outputs flow end-to-end into stats_aggregator_fact with no schema change (they reuse aggregation_label/statistic_value).
• The dimension row's statistics lists all labels. The definition hash folds in each statistic's kind, labels, resolved input indices, params, and function bytecode so changes invalidate cached results, while custom-free aggregators keep their previous hash.

Tests

• unit and integration coverage for single/multi-output, params, input subsetting, empty-interval handling, validation errors, definition-hash stability, and end-to-end fact-row production. Full suite green; black and ruff clean.

Test Plan

• Unit tests added/updated
• Manual testing completed
• Documentation updated (if applicable)

Checklist

• Code follows project style guidelines
• Self-review completed
• No new linter warnings introduced

[codecov]

Codecov Report

❌ Patch coverage is 92.69406% with 16 lines in your changes missing coverage. Please review.
✅ Project coverage is 88.94%. Comparing base (f654d22) to head (39674eb).

Files with missing lines	Patch %	Lines
...ine/analyze/query/aggregations/custom_statistic.py	87.75%	3 Missing and 3 partials ⚠️
...impulse_reporting/aggregations/stats_aggregator.py	91.04%	3 Missing and 3 partials ⚠️
...ine/analyze/query/aggregations/stats_aggregator.py	96.03%	3 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #67      +/-   ##
==========================================
+ Coverage   88.72%   88.94%   +0.22%     
==========================================
  Files          59       60       +1     
  Lines        4826     5013     +187     
  Branches      552      596      +44     
==========================================
+ Hits         4282     4459     +177     
- Misses        445      451       +6     
- Partials       99      103       +4     

Files with missing lines	Coverage Δ
...ngine/analyze/query/aggregations/statistic_type.py	100.00% <100.00%> (ø)
...e_reporting/aggregations/point_value_aggregator.py	97.64% <ø> (ø)
...ine/analyze/query/aggregations/stats_aggregator.py	95.83% <96.03%> (+4.03%)	⬆️
...ine/analyze/query/aggregations/custom_statistic.py	87.75% <87.75%> (ø)
...impulse_reporting/aggregations/stats_aggregator.py	94.17% <91.04%> (-1.54%)	⬇️

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.