Add post-transcript implementation plan

docs/roadmap/post-transcript-artifact-workflow.md

@@ -193,12 +193,14 @@ override them. The shared command parsing layer should reject duplicate
 singleton options rather than use ordering to choose a winner. That rule should
 apply consistently to bounded `run` itself, not only to the alias.
 
-Without `--artifacts`, the alias force-runs extraction and all configured
+Without `--artifacts`, the alias force-runs extraction and all enabled
-analysis artifacts. With `--artifacts`, extraction still produces its complete
+configured analysis artifacts, matching the existing default analysis
+selection. With `--artifacts`, extraction still produces its complete
 configured Notarius bundle, while forced analysis targets only the selected
-Scriptorium artifacts and any prerequisites required to build them. It never
+Scriptorium artifacts and any prerequisites required to build them. An
-runs publish or notify. Changed results may correctly leave those later stages
+explicitly selected disabled artifact remains a valid target under the existing
-stale.
+selection rules. The alias never runs publish or notify. Changed results may
+correctly leave those later stages stale.
 
 The existing `analyze` command remains the convenience path for forcing
 analysis without rerunning Notarius.
@@ -344,60 +346,6 @@ versioned metadata. Readers must accept older manifests, while new writers must
 emit one canonical representation and must not maintain parallel legacy and new
 analysis state indefinitely.
 
-## Required Test Coverage
-
-Allocate tests to the behavior owner and avoid duplicating every case at every
-layer:
-
-- planner and command tests prove the new canonical order, inclusive range
-  selection, omitted-bound defaults, invalid/reversed bounds, duplicate-option
-  rejection, artifact-range validation, and unchanged unbounded behavior;
-- alias tests prove exact argument equivalence and pass-through without
-  re-testing the runner through an independent path;
-- invalidation tests prove the complete fixed relation, inventory validation,
-  deterministic ordering, unknown-stage handling, and render/extract
-  independence;
-- runner tests prove force is confined to the selected range, excluded stages
-  are not resume-validated or composed unnecessarily, outside dependents may be
-  staled but not executed, and unusable prerequisites do not widen the range;
-- continuation tests cover representative manifests produced under the old
-  render/extract order;
-- analyze tests prove fingerprint stability and sensitivity, dependency
-  ordering, selected force, stale-prerequisite rebuilding, valid unselected
-  preservation, changed-dependent invalidation, identical-content reuse,
-  removed configuration, output tampering, partial failure, and legacy-result
-  handling; and
-- catalog and publish tests prove that only current per-artifact results are
-  exposed downstream.
-
-Tests must assert observable plans, manifest transitions, adapter requests, and
-artifact availability rather than private map layout. The default suite remains
-offline, deterministic, and independent of installed Seriatim, Notarius, or
-Scriptorium binaries.
-
-## Documentation And Policy Changes
-
-When implementation lands, update all current-behavior owners in the same
-change:
-
-- revise the architecture policy so canonical order owns execution,
-  dependency-aware invalidation owns affected work, and bounded runs remain
-  contiguous selections rather than arbitrary workflows;
-- document the new order, range flags, exact alias expansion, force scope,
-  plan behavior, incremental reuse, and legacy-analysis transition in the CLI
-  and operations references;
-- update the internal overview and the focused render, extract, analyze,
-  manifest, artifact, adapter, and publish documents where their implemented
-  contracts change;
-- update troubleshooting guidance for incomplete bounded prerequisites, stale
-  analysis artifacts, fingerprint limitations, and forced recovery; and
-- remove every current claim that extract runs between trim and render or that
-  all invalidation is a canonical-order suffix.
-
-Documentation should continue to describe one explicit pipeline, not a DAG
-scheduler. Volatile command syntax and manifest details remain in their
-canonical owners rather than being copied across orientation documents.
-
 ## Compatibility And Operational Effects
 
 - Existing unbounded `run`, `run-stage`, `analyze`, and `publish` commands keep