Release: verify Wesper 0.4.1 compatibility for 0.9.7

[noeltock]

Problem

Block Runner still pins Wesper 0.0.3. Wesper 0.4.1 retains contextVersion 1 while adding optional ownership/taxonomy evidence and more explicit partial warnings. Verify BR consumes those manifests without losing optional data or unnecessarily blocking generation before releasing 0.9.7.

Solution

Pin the published package exactly, add consumer regressions, and exercise BR against a freshly collected site manifest. Keep the existing collection and token-resolution architecture.

 dependencies
-  wesper: 0.0.3
+  wesper: 0.4.1
 context consumption
+  verify optional unknown owners and taxonomy associations survive collection
+  verify partial warnings permit supported generation
+  keep absent block registries non-authoritative
 release
-  block-runner: 0.9.6
+  block-runner: 0.9.7

Read first: src/context/run.ts, src/tokens/context.ts, src/styles/capabilities.ts. Regression coverage: dev/test/context.test.ts, dev/test/styles.test.ts, existing token tests. The older prepared Wesper adapter branch changes REST/WP-CLI collection and is outside this update.

Acceptance criteria

• Manifest and lockfile pin Wesper 0.4.1 without additional runtime dependencies.
• Existing context, token and style tests pass, including optional-field and warning regressions.
• A refreshed WP-CLI manifest drives a representative native-block generation and validation; retained evidence distinguishes collection from browser/rendering proof.
• Full release-candidate CI passes.
• Version and changelog identify 0.9.7 and the already-merged CI improvements.
• Trusted publishing completes its full release matrix; fresh npm consumer and provenance checks pass.

Scope

No ownership-based rendering inference, new WordPress plugin installation, consumer migration, model benchmark or adoption of the older adapter branch. Release evidence must retain warnings and distinguish partial collection from unusable input.

Shipped

PR #108 merged as 42cbd75; 0.9.7 is npm latest with Wesper pinned to 0.4.1. Candidate CI, post-merge CI and the full release workflow passed. Fresh registry installation, context-driven generation/validation and skill installation passed. npm audit signatures verified 367 package signatures and 79 attestations. Optional model benchmarking and private-site browser rendering were not part of this release proof.