feat(host_agent): sparklines + load/core + PSI on the host panel

- Each at-a-glance metric (CPU, Memory, Disk /, Load) now shows a 6h sparkline
  (reused core.status.sparkline_svg + a recent-series query; disk uses the root
  mount sub-resource) so trend/consistency is visible, not just the instant.
- Load is now normalized: "Load /core" = 1m load ÷ CPU cores as % (100% = run
  queue matches capacity), comparable across different hardware. Cores derived
  from the per-core CPU metrics already collected — no agent change. Raw load
  in the tooltip.
- Added a "Pressure 10s" line: PSI cpu/mem/io (some, avg10), the
  hardware-independent saturation signal already collected by the agent.

Scribe #898.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

plugins/host_agent/routes.py

@@ -11,7 +11,7 @@ import secrets
 from quart import Blueprint, current_app, jsonify, request, Response, render_template, redirect, url_for, session
 from steward.auth.middleware import require_role
 from steward.models.users import UserRole
-from sqlalchemy import select, func, or_
+from sqlalchemy import select, func, or_, and_
 from datetime import timedelta
 
 from steward.core.settings import public_base_url
@@ -550,10 +550,47 @@ async def host_panel(host_id: str):
         target = (await db.execute(select(AnsibleTarget).where(
             AnsibleTarget.host_id == host_id))).scalar_one_or_none()
 
+        # Recent series for the at-a-glance sparklines (cpu/mem/load host-level,
+        # disk from the root mount sub-resource).
+        series_raw: dict[str, list[float]] = {"cpu": [], "mem": [], "disk": [], "load": []}
+        if reg:
+            since = datetime.now(timezone.utc) - timedelta(hours=6)
+            spark_rows = (await db.execute(
+                select(PluginMetric).where(
+                    PluginMetric.source_module == SOURCE_MODULE,
+                    PluginMetric.recorded_at >= since,
+                    or_(
+                        and_(PluginMetric.resource_name == host.name,
+                             PluginMetric.metric_name.in_(("cpu_pct", "mem_used_pct", "load_1m"))),
+                        and_(PluginMetric.resource_name == host.name + ":/",
+                             PluginMetric.metric_name == "disk_used_pct"),
+                    ),
+                ).order_by(PluginMetric.recorded_at)
+            )).scalars().all()
+            for r in spark_rows:
+                if r.resource_name == host.name:
+                    key = {"cpu_pct": "cpu", "mem_used_pct": "mem", "load_1m": "load"}.get(r.metric_name)
+                    if key:
+                        series_raw[key].append(r.value)
+                else:
+                    series_raw["disk"].append(r.value)
+
+    from steward.core.status import sparkline_svg
     hostlvl = latest.get(host.name, {})
     # At-a-glance disk = the root filesystem (what people actually care about),
     # not the "worst" mount which is opaque. Worst stays on the full-metrics page.
     disk_root = (latest.get(host.name + ":/", {}) or {}).get("disk_used_pct")
+    sparks = {k: sparkline_svg(v[-120:]) for k, v in series_raw.items()}
+    # Normalize load by core count so it's comparable across hardware (load/core
+    # as %). Cores derived from the per-core CPU sub-resources already collected.
+    cores = sum(1 for k in latest if k.startswith(host.name + ":core"))
+    load1 = hostlvl.get("load_1m")
+    load_per_core = round(load1 / cores * 100) if (cores and load1 is not None) else None
+    psi = {
+        "cpu": hostlvl.get("psi_cpu_some_avg10"),
+        "mem": hostlvl.get("psi_mem_some_avg10"),
+        "io": hostlvl.get("psi_io_some_avg10"),
+    }
     ls = reg.last_seen_at if reg else None
     # "Deployed" is gated on the agent actually checking in (last_seen_at), NOT on
     # the registration row — that row is minted before the playbook runs, so a
@@ -565,6 +602,7 @@ async def host_panel(host_id: str):
     return await render_template(
         "panel.html",
         host=host, reg=reg, hostlvl=hostlvl, disk_root=disk_root,
+        sparks=sparks, cores=cores, load1=load1, load_per_core=load_per_core, psi=psi,
         reporting=reporting, stale=stale, target=target,
         ansible_available=has_capability("ansible.run_playbook"),
         managed_key_set=bool((ansible_cfg.get("ssh_public_key") or "").strip()),