Fix slot boundary alignment for evaluation window end-time log

[Copilot]

last_time in the "Evaluating next N slots until HH:MM" log was computed from the raw calc_timestamp (e.g. 14:23:45) rather than the start of the current slot, producing an off-boundary, misleading time — especially for 15-minute intervals.

Changes

• src/batcontrol/logic/default.py: Replace the display_minutes / t0 / t1 computation with slot-aligned arithmetic:
  • slot_start — floor calc_timestamp to the nearest interval_minutes boundary (zero seconds/microseconds)
  • last_time — slot_start + max_slots * interval_minutes, the true end of the evaluation window

# Before (inaccurate — mid-slot offset propagates into result)
display_minutes = (max_slots * self.interval_minutes) - self.interval_minutes
last_time = (calc_timestamp + timedelta(minutes=display_minutes)).strftime("%H:%M")
# At 14:23:45, 3×15-min slots → 14:53 ✗

# After (aligned to slot boundary)
slot_start = calc_timestamp.replace(
    minute=(calc_timestamp.minute // self.interval_minutes) * self.interval_minutes,
    second=0, microsecond=0
)
last_time = (slot_start + timedelta(minutes=max_slots * self.interval_minutes)).strftime("%H:%M")
# At 14:23:45, 3×15-min slots → 15:00 ✓

---

🔒 GitHub Advanced Security automatically protects Copilot coding agent pull requests. You can protect all pull requests by enabling Advanced Security for your repositories. Learn more about Advanced Security.