fix(analysis): _expected_ticks usa ceiling division (no off-by-one)

Il piano originale aveva `floor(span/15) + 1` che over-conta a span allineati
(span=60min → 5 invece di 4). Il primo fix dell'implementer (`floor(span/15)`)
under-conta a span non-allineati (span=16min → 1 invece di 2). Solo
`ceil(span/15)` è corretto in entrambi i casi. Aggiunti 2 test che
coprono gli scenari non-allineato e boundary-esatto per impedire
regressioni. Plan doc allineato.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

src/cerbero_bite/analysis/data_audit.py

@@ -12,6 +12,7 @@ parameters. To tune a threshold, edit this file.
 
 from __future__ import annotations
 
+import math
 import sqlite3
 import statistics
 from dataclasses import dataclass, field
@@ -117,5 +118,9 @@ def _expected_ticks(since: datetime, until: datetime) -> int:
         )
     if first_tick >= until:
         return 0
-    span = until - first_tick
-    return int(span.total_seconds() // (_TICK_INTERVAL_MIN * 60))
+    # Count ticks in [first_tick, until): the largest k with
+    # first_tick + k*15min < until is ceil(span/15) - 1, so the
+    # count is ceil(span_minutes / 15). floor() under-counts at
+    # aligned multiples and would mis-count non-aligned spans.
+    span_seconds = (until - first_tick).total_seconds()
+    return math.ceil(span_seconds / (_TICK_INTERVAL_MIN * 60))