Fix failed manual run pushing next_run and premature last_run in UI

- record_failure() now only recomputes next_run when the job is already
  overdue (next_run <= now), preserving the scheduled fire time when a
  manual run fails before the job's natural next_run
- Remove premature job.last_run update in scheduler.js — the job runs
  asynchronously so last_run should only reflect the server-side
  completion timestamp on the next data refresh

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

crates/openfang-api/static/js/pages/scheduler.js

@@ -204,7 +204,9 @@ function schedulerPage() {
         var result = await OpenFangAPI.post('/api/cron/jobs/' + job.id + '/run', {});
         if (result.status === 'triggered' || result.status === 'completed') {
           OpenFangToast.success('Job "' + (job.name || 'job') + '" triggered');
-          job.last_run = new Date().toISOString();
+          // Don't update job.last_run here — the job runs asynchronously in the
+          // background. The real last_run is set by the server on completion and
+          // will appear on the next data refresh.
         } else {
           OpenFangToast.error('Run failed: ' + (result.error || 'Unknown error'));
         }

crates/openfang-kernel/src/cron.rs

@@ -387,7 +387,13 @@ impl CronScheduler {
                 );
                 meta.job.enabled = false;
             } else {
-                meta.job.next_run = Some(compute_next_run_after(&meta.job.schedule, Utc::now()));
+                // Only recompute next_run if the job was already overdue. This
+                // preserves the scheduled fire time when a manual (on-demand) run
+                // fails before the job's natural next_run.
+                let now = Utc::now();
+                if meta.job.next_run.map(|t| t <= now).unwrap_or(true) {
+                    meta.job.next_run = Some(compute_next_run_after(&meta.job.schedule, now));
+                }
             }
         }
     }