Author and inspect encounter regions, pack rosters, and offering sites

[zoeyrose]

Outcome

Let authors paint and inspect spawn zones, choose stable rosters/profiles/
policies, lay out offering slots/recipes, and understand static bounds and
diagnostics without manually editing records or reimplementing the authoritative
Go population controller.

This is replacement editor work. Classic editor/ only packages external
Gridarta; Classic authoring compatibility remains in the content schema/tools.

Scope and constraints

• Consume released content-toolkit document, catalog, diagnostics, transaction,
  and compilation APIs. No second parser, validator, serializer, catalog, or
  filesystem write path.
• Paint/select deterministic masks or rectangle unions with undo/redo and
  atomic toolkit transactions; support add/remove/split/merge and clear visual
  distinction between spawn, activation, keepout, leash, protected, and offering
  slot areas.
• Pick stable family/roster/pack/policy/site/recipe/effect references from the
  catalog and show full source-aware linked diagnostics.
• Inspect min/base/max threat and entity bounds, difficulty/player curves,
  refill/cooldown/burst settings, eligible candidates, role constraints, and
  statically admissible compositions.
• Show non-authoritative range/composition previews across selected difficulty
  and one/two/four-player inputs. Label previews as authored bounds; do not copy
  Go scheduling/RNG/placement logic or promise the exact next live pack.
• Lay out semantic left/center/right or named offering slots in the map, edit
  stable item requirements and a bounded typed directive, and preview the
  linked telegraph/reward references.
• Overlay statically invalid or risky geometry, exits/protected cells, impossible
  role constraints, orphan references, overlapping slots, and unsupported local
  overrides. Runtime-only occupancy remains a playtest concern.
• Keep toolkit documents authoritative and preserve external-edit detection,
  revision preconditions, dry-run diff, atomic save, autosave/recovery, and path
  safety.
• Actual gameplay verification launches through the existing isolated playtest
  path; the editor does not embed a server or connect to Classic/Gridarta.

Acceptance criteria

• Authors can create, select, reshape, split, merge, and remove bounded zone
  geometry with undo/redo and exact semantic overlays.
• Every reference picker uses stable catalog IDs and linked diagnostics;
  display names are presentation only.
• Inspector/previews show configured target/entity/threat/role/refill bounds
  for low/normal/high and one/two/four-player inputs without implementing an
  alternate authoritative director.
• Offering slots have deterministic semantic/world positions, recipe item
  references, validation, and typed effect editing with visible cooldown/
  cap constraints.
• Invalid geometry, impossible packs, missing references, unsafe keepouts,
  overlapping slots, and unsupported overrides block publication with
  source-aware diagnostics.
• Opening/saving unchanged authored documents is byte-identical; a targeted
  zone/site edit does not churn unrelated source.
• Stale revision, external edit, validation, conflict, disk/permission,
  cancellation, or crash failure leaves source unchanged and recoverable.
• Headless command/document tests cover geometry operations, reference
  changes, diagnostics, undo/redo, reopen, recovery, and large/malformed
  inputs; visual tests use shared renderer semantic masks.
• One isolated playtest launches the bat/orc/offering pilot through the
  wrapper-owned profile/topology once replacement integration is available.
• tools/validate.sh and git diff --check pass.

Dependencies

Depends on the released toolkit contract and
editor#4,
editor#5, and
editor#6. It does not block the
CLI-authored M3 pilot; it provides the M4 safe authoring workflow.