[FEATURE] Consolidated digest for multiple scheduled Routines firing close together

Summary

Users running multiple scheduled Routines (cron-based recurring agents) across different repos/projects get one separate session and one separate notification per Routine — even when several fire within the same daily window. Please add an optional consolidated digest that rolls same-window Routine firings into a single summary notification/session, linking out to each full transcript.

Current behavior

With several recurring Routines set up across different project directories (e.g. one per repo for a daily health/status check), each firing produces its own session and its own push notification, independently of how close together they run. In practice this means:

• Electure daily check (cwd: ~/CascadeProjects/electure)
• Daily repo review deploy (cwd: ~/CascadeProjects)
• Mamme mart daily check (cwd: ~/CascadeProjects/mamme-mart)
• Driveway daily check (cwd: ~/CascadeProjects/driveway)

...all fire within the same rough morning window (some literally within the same minute, e.g. two routines both timestamped 2026-05-09T23:15:0x), each landing as its own sidebar entry and its own notification. Reviewing "what happened overnight across my projects" means opening N separate sessions and manually piecing together status across all of them.

Motivation

This is a real, ongoing pattern for anyone using Routines to monitor/maintain multiple repos (solo devs, agencies, freelancers managing several client codebases). The more Routines you set up — which is the whole value proposition of the feature — the noisier the notification surface and sidebar become, to the point where the volume of recurring-check overhead starts to offset the time the Routines were meant to save.

Proposed behavior

1. Optional digest grouping. Either a global setting or a per-Routine "digest group" tag lets Routines scheduled within a defined window (e.g. a rolling N-minute window, or simply "same calendar day") get bundled into a single summary notification/session instead of N separate ones.
2. One-line status per Routine inside the digest — success / failure / needs-attention — each linking to that Routine's full session transcript for detail.
3. Individual sessions still exist and are still openable — only the notification/at-a-glance surface consolidates; nothing about the underlying execution changes.
4. Opt-in, not default — users who want per-Routine notifications keep today's behavior; digesting is something you turn on per Routine or per group.

Why it matters

• Cuts notification and sidebar clutter proportionally to how many Routines a user runs — which is exactly when it's most needed.
• Turns "check N sessions every morning" into "glance at one digest, drill in only where something needs attention."
• Makes Routines scale better as a monitoring tool across a growing number of projects, rather than degrading with adoption.

Related issues (adjacent, not duplicates)

• #54517 [FEATURE] Allow scheduled agent / Routine runs to appear in Recents or be pinnable in the sidebar — adjacent (surfacing individual routine runs), different ask (pinning single runs vs. digesting many)
• #48322 [FEATURE] Team/Enterprise: shared routines (org-owned scheduled agents) — adjacent (routine ownership/sharing), unrelated to notification aggregation

Notes

• Grouping should be explicit/opt-in (e.g. a "digest group" field on a Routine's config) rather than inferred purely from timing, so unrelated routines that happen to fire close together aren't silently merged unless the user wants that.