Add overlay runtime import for packed events

docs/README.md

@@ -67,17 +67,18 @@ Current local tool status:
 The atlas milestone is broad enough that the next implementation focus has already shifted downward
 into runtime rehosting. The current runtime baseline now includes deterministic stepping, periodic
 trigger dispatch, normalized runtime effects, staged event-record mutation, fixture execution,
-state-diff tooling, tracked save-slice documents for captured-runtime inputs, and a packed-event
-persistence bridge that now reaches per-record summaries and selective executable import.
+state-diff tooling, tracked save-slice documents for captured-runtime inputs, overlay import
+documents that combine captured snapshots with save-derived state, and a packed-event persistence
+bridge that now reaches per-record summaries and selective executable import.
 
 The highest-value next passes are now:
 
 - preserve the atlas and function map as the source of subsystem boundaries while continuing to
   avoid shell-first implementation bets
+- use captured-context overlay imports whenever save-derived packed rows need live runtime context
+  that the save slice does not actually persist
 - widen packed-event target-family coverage only where static evidence is strong enough to support
-  deterministic executable import
-- add the next imported object/context families needed to turn current parity-only packed rows into
-  executable runtime records
+  deterministic executable import after the necessary runtime context is present
 - use `rrt-hook` primarily as optional capture or integration tooling, not as the first execution
   environment
 - keep `docs/runtime-rehost-plan.md` current as the runtime baseline and next implementation slice

docs/runtime-rehost-plan.md

@@ -22,12 +22,12 @@ Implemented today:
 - snapshots, state dumps, save-slice projection, and normalized state diffing already exist in the
   CLI and fixture layers
 - checked-in runtime fixtures already cover deterministic stepping, periodic service, direct trigger
-  service, snapshot-backed inputs, save-slice-backed inputs, normalized state-fragment assertions,
-  and imported packed-event execution
+  service, snapshot-backed inputs, save-slice-backed inputs, overlay-import-backed inputs,
+  normalized state-fragment assertions, and imported packed-event execution
 
 That means the next implementation work is breadth, not bootstrap. The recommended next slice is
-wider packed-event target-family coverage plus company-collection import depth, not another
-persistence scaffold pass.
+captured-context overlay import for company-targeted packed events, not another persistence
+scaffold pass.
 
 ## Why This Boundary
 
@@ -366,30 +366,32 @@ Checked-in fixture families already include:
 
 ## Next Slice
 
-The recommended next implementation slice is wider packed-event target-family coverage on top of the
-captured save-slice workflow that now exists today.
+The recommended next implementation slice is captured-context overlay import on top of the
+save-slice and snapshot workflows that already exist today.
 
 Target behavior:
 
-- expand the executable import subset beyond the current direct-state and follow-on lanes only when
-  target resolution and field semantics are statically grounded enough to preserve headless
-  determinism
-- add the next imported object families needed to make currently parity-only packed rows executable,
-  starting with company-targeted effects
+- preserve save slices as partial state rather than pretending they reconstruct full live company
+  state
+- overlay save-derived packed-event state onto a captured runtime snapshot that already has the
+  needed company roster and other live context
+- upgrade currently blocked company-targeted packed rows when the overlaid base snapshot provides
+  every referenced company id
 - keep preserving unsupported packed rows as parity summaries instead of guessing executable meaning
 
 Public-model additions for that slice:
 
-- wider target-family summaries only where imported execution can be justified by current static
-  evidence
-- imported company/runtime context needed by the next packed-event target families
-- no shell queue/modal behavior in the runtime core
+- tracked overlay import documents that reference one base snapshot plus one save-slice document
+- runtime-side import outcome labels for packed records so blocked-missing-context and
+  blocked-unsupported cases stay explicit
+- fixture support for generic runtime-import documents, not just snapshots or save slices
 
 Fixture work for that slice:
 
-- save-slice-backed fixtures that prove real packed event records survive import and diff paths
-- regression fixtures that lock the current selective executable import boundary and the
-  unsupported/parity-only counts
+- overlay-backed fixtures that prove company-targeted packed rows execute deterministically against
+  captured company context
+- regression fixtures that lock the before/after boundary between save-slice-only imports and
+  overlay-backed imports
 - state-fragment assertions that lock both packed parity summaries and imported executable records
 
 Do not mix this slice with: