feat(sdk,cli): fill gaps for run items and metadata management

[santi698]

Summary

Fills SDK/CLI gaps for run items and metadata management, per TSSDK-63.

Adds SDK methods and corresponding CLI commands:

SDK method	CLI command	API endpoint
getRunItem(runId, externalId)	runs items get <runId> <externalId>	GET /v1/runs/{run_id}/items/{external_id}
updateRunMetadata(runId, customMetadata)	runs metadata set <runId> <json>	PUT /v1/runs/{run_id}/custom-metadata
updateRunItemMetadata(runId, externalId, customMetadata)	runs items metadata set <runId> <externalId> <json>	PUT /v1/runs/{run_id}/items/{external_id}/custom-metadata
deleteRunResults(runId)	runs results delete <runId>	DELETE /v1/runs/{run_id}/artifacts

--items-style JSON parsing for the metadata commands supports "null" to explicitly clear existing custom metadata.

Side fix

Found and fixed a stale nested packages/cli/node_modules/@aignostics/sdk@3.6.1 (a real directory, not a symlink) that shadowed the workspace symlink to packages/sdk. It silently broke CLI tests exercising new SDK methods (sdk.updateRunMetadata is not a function) even though the SDK source/dist had them. A full clean reinstall (rm -rf node_modules packages/*/node_modules package-lock.json && npm install) regenerated a correct lockfile — this also reduced npm audit findings on this branch from 7 to 1 (low severity).

Testing

• npm run test (both packages) — 251 tests passing (55 SDK + 196 CLI)
• npm run typecheck — clean
• npm run lint — clean
• npm run build — clean

Refs: TSSDK-63

[sonarqubecloud]

Quality Gate failed

Failed conditions
0.0% Coverage on New Code (required ≥ 70%)

See analysis details on SonarQube Cloud