Fix recurring task logging as if it enqueued the same job twice

[ghiculescu]

A bug exists with recurring task enqueuing, where you will sometimes see doubled up [ActiveJob] Enqueued ... log lines. On Postgres these do not result in doubled up enqueueing, but they do result in confusing log lines. For example, given a QueueHeartbeatJob that's meant to run every minute:

2026-04-09T06:50:00Z app[e823931b293218] syd [info][ActiveJob] Enqueued QueueHeartbeatJob (Job ID: f3f21750-2d6b-416e-a1f3-5a1235744825) to SolidQueue(default)
2026-04-09T06:50:00Z app[e823931b293218] syd [info][ActiveJob] [QueueHeartbeatJob] [f3f21750-2d6b-416e-a1f3-5a1235744825] Performing QueueHeartbeatJob (Job ID: f3f21750-2d6b-416e-a1f3-5a1235744825) from SolidQueue(default) enqueued at 2026-04-09T06:50:00.002768396Z
2026-04-09T06:50:00Z app[e823931b293218] syd [info][ActiveJob] [QueueHeartbeatJob] [f3f21750-2d6b-416e-a1f3-5a1235744825] Performed QueueHeartbeatJob (Job ID: f3f21750-2d6b-416e-a1f3-5a1235744825) from SolidQueue(default) in 11.79ms
2026-04-09T06:51:00Z app[e823931b293218] syd [info][ActiveJob] Enqueued QueueHeartbeatJob (Job ID: e32694c9-eeb8-4079-bc7c-6c77a405fd6f) to SolidQueue(default)
2026-04-09T06:51:00Z app[e823931b293218] syd [info][ActiveJob] [QueueHeartbeatJob] [e32694c9-eeb8-4079-bc7c-6c77a405fd6f] Performing QueueHeartbeatJob (Job ID: e32694c9-eeb8-4079-bc7c-6c77a405fd6f) from SolidQueue(default) enqueued at 2026-04-09T06:51:00.001187865Z
2026-04-09T06:51:00Z app[e823931b293218] syd [info][ActiveJob] [QueueHeartbeatJob] [e32694c9-eeb8-4079-bc7c-6c77a405fd6f] Performed QueueHeartbeatJob (Job ID: e32694c9-eeb8-4079-bc7c-6c77a405fd6f) from SolidQueue(default) in 9.0ms
2026-04-09T06:52:00Z app[e823931b293218] syd [info][ActiveJob] Enqueued QueueHeartbeatJob (Job ID: e60a1384-4b46-4c97-94bc-7dd695d1cfdb) to SolidQueue(default)
2026-04-09T06:53:00Z app[e823931b293218] syd [info][ActiveJob] Enqueued QueueHeartbeatJob (Job ID: d5f795de-080c-47bd-88e2-0e33520aa470) to SolidQueue(default)
2026-04-09T06:53:00Z app[e823931b293218] syd [info][ActiveJob] [QueueHeartbeatJob] [d5f795de-080c-47bd-88e2-0e33520aa470] Performing QueueHeartbeatJob (Job ID: d5f795de-080c-47bd-88e2-0e33520aa470) from SolidQueue(default) enqueued at 2026-04-09T06:53:00.009200319Z
2026-04-09T06:53:00Z app[e823931b293218] syd [info][ActiveJob] [QueueHeartbeatJob] [d5f795de-080c-47bd-88e2-0e33520aa470] Performed QueueHeartbeatJob (Job ID: d5f795de-080c-47bd-88e2-0e33520aa470) from SolidQueue(default) in 15.97ms

To find the root cause, I deployed SQ with extra logging. If we include those logs and remove the job being performed it makes it easier to see what's going on:

2026-04-09T06:50:00Z app[e823931b293218] syd [info]schedule: QueueHeartbeatJob.perform_later() [ * * * * * ], delay = 59.998386254, next time = 2026-04-09 06:51:00 UTC (now = 2026-04-09 06:50:00 UTC, diff = 59.999722945)
2026-04-09T06:50:00Z app[e823931b293218] syd [info][ActiveJob] Enqueued QueueHeartbeatJob (Job ID: f3f21750-2d6b-416e-a1f3-5a1235744825) to SolidQueue(default)
2026-04-09T06:50:00Z app[e823931b293218] syd [info]SolidQueue-1.4.0 Enqueued recurring task (17.0ms)  task: "queue_heartbeat", active_job_id: "f3f21750-2d6b-416e-a1f3-5a1235744825", at: "2026-04-09T06:50:00Z"
2026-04-09T06:51:00Z app[e823931b293218] syd [info]schedule: QueueHeartbeatJob.perform_later() [ * * * * * ], delay = 59.999577626, next time = 2026-04-09 06:51:00 UTC (now = 2026-04-09 06:50:59 UTC, diff = 8.0566e-05)
2026-04-09T06:51:00Z app[e823931b293218] syd [info][ActiveJob] Enqueued QueueHeartbeatJob (Job ID: e32694c9-eeb8-4079-bc7c-6c77a405fd6f) to SolidQueue(default)
2026-04-09T06:51:00Z app[e823931b293218] syd [info]SolidQueue-1.4.0 Enqueued recurring task (19.1ms)  task: "queue_heartbeat", active_job_id: "e32694c9-eeb8-4079-bc7c-6c77a405fd6f", at: "2026-04-09T06:51:00Z"
2026-04-09T06:52:00Z app[e823931b293218] syd [info]schedule: QueueHeartbeatJob.perform_later() [ * * * * * ], delay = 59.99585098, next time = 2026-04-09 06:53:00 UTC (now = 2026-04-09 06:52:00 UTC, diff = 59.999882842)
2026-04-09T06:52:00Z app[e823931b293218] syd [info][ActiveJob] Enqueued QueueHeartbeatJob (Job ID: e60a1384-4b46-4c97-94bc-7dd695d1cfdb) to SolidQueue(default)
2026-04-09T06:52:00Z app[e823931b293218] syd [info]SolidQueue-1.4.0 Skipped recurring task – already dispatched (20.4ms)  task: "queue_heartbeat", at: "2026-04-09T06:51:00Z"
2026-04-09T06:53:00Z app[e823931b293218] syd [info]schedule: QueueHeartbeatJob.perform_later() [ * * * * * ], delay = 59.993155042, next time = 2026-04-09 06:54:00 UTC (now = 2026-04-09 06:52:59 UTC, diff = 60.000016247)
2026-04-09T06:53:00Z app[e823931b293218] syd [info][ActiveJob] Enqueued QueueHeartbeatJob (Job ID: d5f795de-080c-47bd-88e2-0e33520aa470) to SolidQueue(default)
2026-04-09T06:53:00Z app[e823931b293218] syd [info]SolidQueue-1.4.0 Enqueued recurring task (16.4ms)  task: "queue_heartbeat", active_job_id: "d5f795de-080c-47bd-88e2-0e33520aa470", at: "2026-04-09T06:53:00Z"

On line 1, next time = 2026-04-09 06:51:00 UTC - that's when we expect the next Concurrent::ScheduledTask to fire. On line 4 we see it's actually fired at 2026-04-09 06:50:59 UTC. As a result, the new next_time is incorrectly calculated as the current time. A job has already been enqueued for 06:51 so the second attempt will result in a uniqueness validation - but not before the [ActiveJob] Enqueued QueueHeartbeatJob line is logged.

This PR fixes this by calculating next_time in terms of when we thought the Concurrent::ScheduledTask would fire, not the actual time. This handles the case where it executes milliseconds earlier than we expected to.