metrics: guard against corrupted time streams causing OOM

Strava originals with absolute Unix timestamps stored as elapsed-second
offsets produce a t_max of ~1.6 billion. compute_mmp and compute_best_efforts
both create dense 1Hz arrays via range(t_min, t_max+1), which for a 1.6B
span allocates 44+ GB and OOM-kills the process. Add a >1-week sanity
check and return None early for corrupt streams.

Root cause: old Strava activities (seen from 1970-epoch start_date)
where the time stream contains absolute Unix timestamps instead of
elapsed seconds.

bincio/extract/metrics.py

@@ -131,6 +131,10 @@ def compute_mmp(pts: list[DataPoint], started_at: datetime) -> Optional[list[lis
 
     t_min = min(sparse)
     t_max = max(sparse)
+    # Guard against corrupted time data (e.g. absolute Unix timestamps stored as
+    # elapsed offsets, which can make t_max astronomically large and OOM the process).
+    if t_max - t_min > 7 * 24 * 3600:  # > 1 week → corrupted stream
+        return None
     power_1hz: list[int] = [sparse.get(t, 0) for t in range(t_min, t_max + 1)]
 
     n = len(power_1hz)
@@ -190,6 +194,10 @@ def compute_best_efforts(
 
     t_min = min(sparse_speed)
     t_max = max(sparse_speed)
+    # Guard against corrupted time data (e.g. absolute Unix timestamps stored as
+    # elapsed offsets, which can make t_max astronomically large and OOM the process).
+    if t_max - t_min > 7 * 24 * 3600:  # > 1 week → corrupted stream
+        return None, None
     speed_1hz: list[float] = [sparse_speed.get(t, 0.0) for t in range(t_min, t_max + 1)]
     ele_1hz: list[Optional[float]] = [sparse_ele.get(t) for t in range(t_min, t_max + 1)]