fix(analytics): fix active users endpoint SQL type casting

[TheMorpheus407]

Summary

• Fixed PostgreSQL type casting issues in the active users analytics endpoint
• Added explicit timestamptz cast for start parameter
• Added bigint cast for bucket_count parameter
• Changed interval parameter to use integer concatenation instead of string

Details

The /analytics/active-users endpoint was returning "Internal server error" due to SQL type casting issues in the PostgreSQL query. The fixes ensure proper type coercion:

1. Start parameter: Now explicitly casts to timestamptz
2. Bucket count: Now explicitly casts to bigint
3. Interval: Changed from passing a formatted string to passing bucket_seconds as integer and constructing the interval in SQL

Test plan

• Endpoint returns 200 OK with admin authentication
• Returns correct JSON structure with bucket_start and active_users
• Works with different time ranges (1d, 7d, 30d, 90d)
• Works with different granularities (1h, 1d, 7d, 30d)
• Shows actual active user counts from session data
• Tested via curl with admin token

Example Response

[
  {
    "bucket_start": 1761465600,
    "active_users": 1
  },
  {
    "bucket_start": 1761469200,
    "active_users": 2
  }
]

The endpoint now correctly returns active user statistics bucketed by configurable time ranges and granularities for the admin dashboard.

🤖 Generated with Claude Code