fix(rq): replace removed Job.get_id() with Job.id property

[gferrate]

Overview

Fixes an `AttributeError` in the RQ integration that occurs when using RQ 2.x. The `Job.get_id()` method was removed in RQ 2.0; the integration was calling it in three places when tagging spans with the job ID.

Motivation

RQ 2.0 was released in October 2024 and RQ 2.7.0 (the current latest) was released in February 2026. Any user who has upgraded RQ to 2.x will get an `AttributeError` on every traced job enqueue or execution, breaking their instrumentation entirely.

Change

Replace all three `job.get_id()` calls with `job.id`. The `id` property has been available since RQ 1.8.1 (our minimum supported version) — in 1.x it was literally `id = property(get_id, set_id)`, so this change is fully backwards compatible across the supported range (1.8.1 → 2.7.0).

Also updates `registry.yaml` to reflect the tested max version as 2.7.0.

Tests

• Added `test_job_id_tag_on_enqueue_span`: explicitly asserts `job.id` tag on the enqueue span matches the job's actual ID.
• Added `test_job_id_tag_on_worker_spans`: asserts `job.id` on both `rq.worker.perform_job` and `rq.job.perform` spans.
• Added `test_custom_job_id_in_span_tags`: verifies a user-supplied job ID flows through to all span tags.

The existing snapshot tests ignored `meta.job.id`, so they would not have caught this at the assertion level — only at the `AttributeError` level when running against RQ 2.x.

Risks

• Low. The change is a drop-in replacement; `job.id` and `job.get_id()` were identical in 1.x. No behaviour change is expected.
• Potential secondary issue (not in this PR): `test_sync_worker_ttl` calls `job.get_status()` on a job whose Redis data has been deleted (`result_ttl=0`). In RQ 2.x `get_status()` may raise `InvalidJobOperation` instead of returning `None`. This is a test-only issue and does not affect the integration code.

Backport

This is a critical fix for all users on RQ 2.x. Backport to the most recent release branch is recommended.