perf(cron): batch advance_next_run for the due-dispatch loop

[spfcraze]

What does this PR do?

The scheduler's pre-dispatch loop called advance_next_run per due job — one full load_jobs() + one full save_jobs() of the jobs file each — so N due jobs cost N reads + N writes of the whole file (gateway-restart catch-up or co-scheduled bursts). Add advance_next_runs(): one load + at most one save for the whole due set with identical per-job semantics; advance_next_run() becomes a thin wrapper.

Related Issue

No direct issue — discovered via code review and reproduced live (see below).
Related PRs reviewed during the duplicate check (none covers this change):

• fix(cron): deduplicate cross-profile jobs in GET /api/cron/jobs NousResearch/hermes-agent#51742 [open] fix(cron): deduplicate cross-profile jobs in GET /api/cron/jobs
• fix(cron): keep scheduler jobs outside child context NousResearch/hermes-agent#74600 [open] fix(cron): keep scheduler jobs outside child context
• feat: add record_session option to cron jobs NousResearch/hermes-agent#71546 [open] feat: add record_session option to cron jobs
• fix: block persistent gateway self-restart jobs NousResearch/hermes-agent#63272 [open] fix: block persistent gateway self-restart jobs
• fix(cron): hold _jobs_lock() in snapshot-restore paths that overwrite cron/jobs.json NousResearch/hermes-agent#46765 [open] fix(cron): hold _jobs_lock() in snapshot-restore paths that overwrite cron/jobs.json
• ci: add timeout-minutes to workflow jobs NousResearch/hermes-agent#62102 [open] ci: add timeout-minutes to workflow jobs
• fix(cron): let briefing jobs require delivery NousResearch/hermes-agent#53252 [open] fix(cron): let briefing jobs require delivery
• fix(cron): parallelize script jobs with workdirs NousResearch/hermes-agent#61774 [open] fix(cron): parallelize script jobs with workdirs
• …and 65 more reviewed in the sweep

Changes Made

• fix/cron-batch-advance — 3 file(s) changed vs base:
  • cron/jobs.py
  • cron/scheduler.py
  • tests/cron/test_jobs.py

cron/jobs.py: advance_next_runs() batch primitive (identical skip semantics for one-shots/unknown ids, no save when nothing advances; crash-window tradeoff documented); advance_next_run() is now a thin wrapper preserving its exact contract. cron/scheduler.py: the pre-dispatch loop is one batch call. tests/cron/test_jobs.py: +4 tests — batch semantics (recurring advanced, one-shots skipped), single load+save I/O pin, no-save-when-noop, wrapper contract. Related: NousResearch#60946 and NousResearch#75833 restructure this call site for correctness — neither addresses the I/O cost; the batch primitive composes with either dispatch design.

How to Test

Measured on a real jobs file with 50 due recurring jobs (repo venv): the per-job loop costs 107.9 ms (50 load_jobs + 50 save_jobs); the batch call costs 2.5 ms (1 + 1) — 45x. Burst-only path: gateway-restart catch-up or many co-scheduled jobs firing in one tick.

Validation completed (recorded by prp):

1. Sabotage check: pre-fix code fails the regression tests (3 failed), with the fix all pass (58 passed, 0 failed) — target tests/cron/test_jobs.py.
2. Suite tests/cron/: branch 377 passed / 0 failed vs baseline 373 passed / 0 failed — zero branch-only failures.
3. Duplicate check: 73 potential matches reviewed — none covers this change.
4. The full repo-wide suite was not run for this change; GitHub CI owns full-suite validation.

Logs

Sabotage verification output:

# base leg (pre-fix code + branch tests):
#   tests: 55 passed, 3 failed
# head leg (with fix):
#   tests: 58 passed, 0 failed