Mirrors ruby-sdk's img.shields.io/gem/v/...svg pattern. The bare URL key
had a poisoned shields.io negative cache from before first publish; the
.svg key is fresh and renders v1.0.0 immediately. [skip ci] so no CI run
fires and the workflow_run release trigger stays dormant.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Relocates the SDK testing guide into the repo as the canonical source.
Wiki cross-references are rewritten to absolute wiki URLs so they resolve
from the repo file. No code change; docs-only, so semantic-release publishes
no release.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…king

Story 2.5. Tests for the additive bucketing-event delivery path and keyword-only
enable_tracking control. Implementation follows (GREEN).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…call enable_tracking control

Closes the cross-SDK exposure-tracking parity gap (Story 2.5). A visitor that
buckets into a variation now produces a standalone bucketing/activation event
delivered to the metrics tracking endpoint, batched alongside conversion events
in the same events array, so experiment exposure is recorded instead of only
conversions.

Strictly additive: run_experience / run_experiences gain a keyword-only
enable_tracking=True parameter; the ExperienceResult return value and all
local-evaluation behavior are unchanged. Bucketing events dedup by
(visitor_id, experience_id) via the existing DataStore boundary and flow through
the existing Tracker queue/transport and the LifecycleEvent.BUCKETING EventBus.

Wire shape: {"eventType":"bucketing","data":{"experienceId","variationId"}}
(JS VisitorTrackingEvents/BucketingEvent parity; no timestamp).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Code-review follow-up (Story 2.5). Extract the event-construction + payload
extraction into _serialize_event and reuse a single parametrize decorator so the
parity tests stay under the SonarCloud new-code-duplication gate. No behavior
change; 41 bucketing tests + full suite (857) green.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Empty commit to fire pull_request:synchronize so CodeQL default setup produces an
ingested analysis for this PR (its original analysis predated code scanning being
enabled, leaving the code_scanning merge-protection rule with no PR results to evaluate).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…s-parity

fix: emit experience-activation (bucketing) tracking events with per-call enable_tracking control