feat(api): 个人资料/单位/同步偏好 + 运动详情 + 身体年龄

设置 (services/settings.py, routes/settings.py)
- user_settings 表：身高/体重/出生日期/性别/单位/自动同步开关/同步频率/历史范围
- GET|PUT /api/settings，GET /api/settings/options（取值由后端给，前端不臆造）
- GET /api/settings/rating-basis：把每条参考区间的来源公开出来。
  一个把数字标成「偏低」的区间是在下判断，用户有权看到依据。

运动详情 (services/garmin.py)
- GET /api/garmin/activities/<id>/detail：概览/分段/心率区间/天气/装备/采样曲线
- 首次打开回源 Garmin 并落库，之后走缓存；?refresh=1 强制刷新
- 采样点在写入时抽稀到 300，手机图表画不了更多，也免得整行撑大

身体年龄 (services/fitness_age.py)
- 0.2.8 版 garminconnect 没有 fitnessage 接口，改为本地按公开常模推算：
  VO₂max 对应年龄为基准，静息心率与 BMI 做有上限的修正
- 返回每一步的中间值，界面照实展示，不做成一个不可追溯的分数
- 高于参考表最年轻一档时按 20 岁计——那里外推会得到「11 岁」这种结果

调度器改为每 5 分钟 tick，是否该同步按各账号自己的频率判断

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>

backend/services/scheduler.py

@@ -23,6 +23,7 @@ import time
 from config import DB_TYPE
 from db import execute, query_one, query_all
 from services import garmin as garmin_svc
+from services import settings as settings_svc
 
 JOB_NAME = "garmin_auto_sync"
 
@@ -32,6 +33,11 @@ INTERVAL_SECONDS = int(os.environ.get("AUTO_SYNC_INTERVAL_SECONDS") or 3600)
 SYNC_DAYS = int(os.environ.get("AUTO_SYNC_DAYS") or 2)
 ENABLED = (os.environ.get("AUTO_SYNC") or "true").lower() not in ("0", "false", "no")
 
+# The loop wakes on this cadence; whether an account is actually due is then
+# decided per account from its own 同步频率 setting. A single global interval
+# would mean one user's choice of 30 minutes silently applied to everyone.
+TICK_SECONDS = 300
+
 # A claim older than this is treated as abandoned — the worker holding it died
 # mid-run, and without expiry the job would never run again.
 CLAIM_TIMEOUT_SECONDS = 1800
@@ -99,14 +105,45 @@ def release(name=JOB_NAME, ran=True):
         execute("UPDATE job_locks SET claimed_at = NULL WHERE name = ?", [name])
 
 
-def sync_all_accounts(days=None):
-    """Sync every account that has a stored token. Returns a per-user result."""
+def due_at(user_id):
+    """When this account may next be synced automatically, or None if never.
+
+    None means auto-sync is switched off for them; a time in the past means
+    they are due now.
+    """
+    prefs = settings_svc.get_raw(user_id)
+    if not prefs.get("auto_sync"):
+        return None
+    minutes = prefs.get("auto_sync_minutes") or (INTERVAL_SECONDS // 60)
+    last = _parse((garmin_svc.get_sync_status(user_id) or {}).get("lastSyncTime"))
+    if not last:
+        return _now() - datetime.timedelta(seconds=1)
+    return last + datetime.timedelta(minutes=minutes)
+
+
+def sync_all_accounts(days=None, respect_schedule=False):
+    """Sync every account that has a stored token. Returns a per-user result.
+
+    `respect_schedule` is what the background loop passes: it skips accounts
+    that have auto-sync off or that were synced recently enough. A direct call
+    (a manual "sync everything") leaves it False and syncs unconditionally.
+    """
     days = days or SYNC_DAYS
     rows = query_all("SELECT user_id FROM garmin_tokens")
     results = []
     for row in rows:
         uid = row["user_id"]
         try:
+            if respect_schedule:
+                due = due_at(uid)
+                if due is None:
+                    results.append({"user": uid, "status": "skipped",
+                                    "reason": "auto-sync off"})
+                    continue
+                if due > _now():
+                    results.append({"user": uid, "status": "skipped",
+                                    "reason": "not due"})
+                    continue
             out = garmin_svc.sync_data(uid, {}, days=days)
             results.append({"user": uid, "status": out.get("status"),
                             "records": out.get("recordsSynced")})
@@ -118,16 +155,16 @@ def sync_all_accounts(days=None):
 def _loop():
     while True:
         try:
-            if claim():
+            if claim(interval=TICK_SECONDS):
                 try:
-                    sync_all_accounts()
+                    sync_all_accounts(respect_schedule=True)
                 finally:
                     release()
         except Exception as e:  # noqa: BLE001 - the loop must outlive any single failure
             print(f"[scheduler] tick failed: {e}")
         # Checked more often than the interval so a worker that starts late
         # still picks the job up promptly rather than waiting a full hour.
-        time.sleep(min(300, INTERVAL_SECONDS))
+        time.sleep(TICK_SECONDS)
 
 
 def start():
@@ -144,14 +181,27 @@ def start():
     print(f"[scheduler] auto-sync every {INTERVAL_SECONDS}s, {SYNC_DAYS} day(s) back")
 
 
-def status():
+def status(user_id=None):
+    """Scheduler state, and — when a user is named — their own next due time."""
     row = query_one("SELECT * FROM job_locks WHERE name = ?", [JOB_NAME])
     last = _parse(row.get("last_run_at")) if row else None
-    return {
+
+    out = {
         "enabled": ENABLED,
         "intervalSeconds": INTERVAL_SECONDS,
+        "tickSeconds": TICK_SECONDS,
         "days": SYNC_DAYS,
         "lastRunAt": _iso(last) if last else None,
-        "nextRunAt": _iso(last + datetime.timedelta(seconds=INTERVAL_SECONDS)) if last else None,
+        "nextRunAt": _iso(last + datetime.timedelta(seconds=TICK_SECONDS)) if last else None,
         "running": bool(row and row.get("claimed_at")),
     }
+
+    if user_id:
+        prefs = settings_svc.get_raw(user_id)
+        due = due_at(user_id)
+        out["account"] = {
+            "autoSync": bool(prefs.get("auto_sync")),
+            "intervalMinutes": prefs.get("auto_sync_minutes"),
+            "dueAt": _iso(due) if due else None,
+        }
+    return out