fix(sensenova): clamp unsupported max reasoning effort to xhigh

[InkshadeWoods]

Summary

• Normalize reasoning_effort: "max" to xhigh for SenseNova's deepseek-v4-flash, whose upstream contract explicitly accepts up to xhigh but rejects max.
• Preserve literal max passthrough for models without explicit per-model effort metadata and avoid cross-provider capability lookup for same-ID models.
• Add a regression test for the SenseNova capability ceiling.

Related Issues

• Closes #
• Related to #

Validation

Choose the change type and focused loop from the
Contribution Golden Path. The full unit suite,
Vitest, the 60% coverage gate, and the production build all run in CI on this PR (#8329):

• Change type: provider
• Focused tests and category gates from the golden path
  • node --import tsx/esm --test tests/unit/base-reasoning-effort-split.test.ts tests/unit/sensenova-reasoning-effort.test.ts tests/unit/opencode-zen-reasoning-effort.test.ts
  • Result: 12 passed, 0 failed.
  • Manual live validation: input max was normalized to xhigh; sensenova/deepseek-v4-flash returned HTTP 200 with a non-empty reply.
• npm run lint
  • The full local ESLint process hung without diagnostics and was stopped. Targeted ESLint for all changed files exited successfully with code 0.
• Reconciled with the current active release base; focused checks rerun afterward
• Production-code changes include a new or updated automated test in this PR
• SonarQube is temporarily opt-in while the private project has no quota; it is not a PR gate.

Tests Added Or Updated

• Added tests/unit/sensenova-reasoning-effort.test.ts.

Coverage Notes

• tests/unit/sensenova-reasoning-effort.test.ts covers the explicit SenseNova max to xhigh normalization and verifies that a SenseNova model without explicit effort metadata continues to preserve max.
• tests/unit/opencode-zen-reasoning-effort.test.ts confirms same-ID OpenCode models retain their existing max passthrough behavior.
• tests/unit/base-reasoning-effort-split.test.ts confirms both supported import paths resolve to the same sanitizer.
• No known coverage decrease in touched files.

Reviewer Notes

• No migrations or feature flags.
• The capability lookup intentionally stays scoped to the current provider because providers can expose the same model ID with different upstream reasoning-effort contracts.
• The change only affects models that explicitly declare supported reasoning efforts and omit max; models with absent metadata keep the existing passthrough behavior.