Expand projection drift diagnostics

[chrisbliss18]

Why

This improves the rollout projection-drift tooling so Systems operators get actionable diagnostics when the legacy jetpack_monitor_sites.site_status projection disagrees with authoritative v2 event state. Projection drift is a rollout blocker, and a count-only failure does not give enough context to decide whether the issue is isolated, range-wide, stale legacy state, a missing event-to-projection write, or an unexpected event shape.

What changed

• Updates projection-drift SQL to compare each active site row against a per-site rollup of open HTTP events.
• Keeps the query scoped to the requested rollout bucket range so per-host checks do not aggregate unrelated open events.
• Adds bucket/status summary output before individual drift rows.
• Adds likely-cause labels and operator guidance for common mismatch classes.
• Adds an explicit manual-review warning and next-step guidance before the detailed tables.
• Uses all summary groups for cause totals while displaying only the top summary rows, so hidden groups do not make the guidance incomplete.
• Reports hidden summary groups/rows and warns if drift changes while the report is running.
• Sanitizes database-backed table strings before printing them to the terminal.
• Documents that the command is read-only and that any future repair planner should be based on observed rehearsal or production drift classes.

Example output

INFO projection_drift_range=0-99
INFO legacy_projection_drift=3
WARN legacy_projection_drift_requires_manual_review=3
WARN projection_drift_next_step="review the summary first, then inspect listed event rows before making any site_status repair"

## projection drift summary
BUCKET   COUNT   SITE_STATUS            EXPECTED               OPEN_EVENTS SAMPLE_BLOG  CAUSE                               EVENT_STATE
7        2       1/SITE_RUNNING         2/SITE_CONFIRMED_DOWN  1           42           missing_confirmed_down_projection   Down
8        1       0/SITE_DOWN            1/SITE_RUNNING         0           43           stale_legacy_down_projection        -
WARN projection_drift_cause=missing_confirmed_down_projection count=2 action="an open Down event exists but the legacy site row still reports running; inspect the eventstore transaction path before continuing rollout"
WARN projection_drift_cause=stale_legacy_down_projection count=1 action="the legacy site row still reports downtime even though no open HTTP downtime event requires it; inspect recent close transitions before setting the projection back to running"

## projection drift rows
BLOG_ID      BUCKET   SITE_STATUS            EXPECTED               EVENT_ID   OPEN_EVENTS CAUSE                               EVENT_STATE
42           7        1/SITE_RUNNING         2/SITE_CONFIRMED_DOWN  123        1           missing_confirmed_down_projection   Down
INFO projection_drift_rows_truncated=2
INFO projection_drift_repair=manual_confirmation_required
INFO projection_drift_repair_guidance=confirm the authoritative event rows first; then repair the legacy site_status projection inside a reviewed DB change or by rerunning the code path that writes the event and projection together
FAIL legacy projection drift=3 in range 0-99

Validation

• go test ./cmd/jetmon2 ./internal/db
• make rollout-docs-verify