get_day_summary: add include_patterns option

[Bhumi701]

Fixes #17

Adds an optional include_patterns flag (with pattern_days) to
get_day_summary that appends the same repeated-workflow counts as
get_patterns (kind/label/count only, no prose) — so an agent wanting
both no longer needs two tool calls. Defaults to false, so existing
callers see no change.

Scoped per the discussion in the issue thread — no new tool, patterns
stay Tier-1 (measured), and the "main activity in..." sentence-style
summary stays out (that's a salience judgment for the consuming
agent, per SPEC section 7).

Tests: added test_tool_call_get_day_summary_include_patterns in
tests/test_mcp.py. Full suite passes (106 tests).

Checklist

• Output stays deterministic (no clock, network, randomness, or model calls)
• New behavior has a test; pytest -q passes
• New schema fields, if any, are measured - or live in the tier-2 inferred block with a confidence tag

[nossa-y]

Scope matches #17 and default-off is byte-identical to main. Before merge:

1. Anchor the window to the requested day (end = day's window end, start = end - pattern_days) - right now a past day's summary appends patterns from [now - N, now].
2. Make the test assert real content (it's [] on the fixture, so the loop never runs) - with the day anchor it becomes non-empty and clock-independent.
3. Put ### Added under a ## [Unreleased] heading.
4. Match sibling indentation in TOOLS.

[Bhumi701]

Pushed fixes for all 4 review points:

1. Patterns window is now anchored to the requested day (end = day's
   window end, start = end - pattern_days), not "now".
2. Test now asserts real content (linkedin.com/feed pattern) instead of
   an empty list that let the loop silently pass.
3. CHANGELOG: ### Added is under ## [Unreleased], merged cleanly with
   the Fixed/Changed sections.
4. TOOLS docstring and TOOLS list indentation now matches siblings.

Also merged latest main to pick up the force-push. 106 tests pass.