feat: implement host baseline learning (Task 12)

[itenev]

Summary
Implements host baseline learning (Task 12) by computing p50/p95 percentiles from Prometheus historical data and storing them in the SQLite database. This provides Hermes with statistical context for evaluating whether current metric values are anomalous for a specific host.

Type of Change
- [ ] Bug fix (non-breaking change which fixes an issue)
- [x] New feature (non-breaking change which adds functionality)
- [ ] Breaking change (fix or feature that would cause existing functionality to not work as expected)
- [x] Documentation update

Pre-Review Findings and Resolutions
<!-- Hermes pre-review checklist. Fill this out before requesting human review. -->
- [x] Linting/formatting (ruff check --fix && ruff format) applied before pre-review.
- [x] Any runbook loading/YAML changes include an end-to-end test loading the physical file. (N/A for this PR)
- [x] Design contracts verified (single-writer boundary maintained via SQLite ON CONFLICT upsert, no new violations introduced).
- Findings flagged by Hermes: 
  - Initial implementation incorrectly used "value" (singular) instead of "values" (plural) when parsing the Prometheus /api/v1/query_range matrix response, which would have silently returned empty lists in production.
- Resolutions applied:
  - Changed key to "values" and added try/except for ValueError/TypeError during float parsing.
  - Added test_query_prometheus_range_parses_matrix_response to catch this specific regression by feeding a realistic Prometheus JSON response structure through the parser.

Testing
- [x] All existing tests pass (pytest -v --cov=agentic_node_ops --cov=webhook_receiver — 136 tests passed, 89% coverage)
- [x] New tests added for new functionality (tests/test_baselines.py includes 7 tests covering percentile math and realistic Prometheus matrix response parsing)
- [x] Manual verification steps (if applicable): N/A (fully covered by unit tests using mocked Prometheus range responses and temporary in-memory SQLite databases)

Checklist
- [x] My code follows the style guidelines of this project
- [x] I have performed a self-review of my own code
- [x] I have commented my code, particularly in hard-to-understand areas
- [x] I have made corresponding changes to the documentation (docs/hermes-implementation-plan.md)
- [x] My changes generate no new warnings or errors