Let kernel own default telemetry enabled state

[jay-xiao446]

Summary

• pass through explicit enable_telemetry values to the kernel
• omit telemetry_enabled from kernel Session kwargs when the user does not specify enable_telemetry
• update tests to assert the unset/None case lets the kernel default apply

Validation

• python3.12 -m py_compile src/databricks/sql/backend/kernel/client.py src/databricks/sql/session.py tests/unit/test_session.py tests/unit/test_telemetry.py tests/unit/test_kernel_client.py
• git diff --check

[peco-review-bot]

Verdict: 1 Medium

Looks sound — the None-passthrough is reachable (client.py never defaults enable_telemetry into kwargs before Session), the is not None guard correctly forwards explicit True/False while omitting unset, and tests are updated to match. One medium concern: this removes a deliberately-documented opt-out default on the kernel path, and the resulting default now depends on an external databricks_sql_kernel value not visible in this repo — worth confirming it matches intent.

[peco-review-bot]

🟡 Medium — This reverses a previously-deliberate divergence. The old code (and its removed comment) explicitly forced kernel-path telemetry OFF by default — "telemetry_enabled": bool(options.get("enable_telemetry", False)) in client.py — and warned "Do not 'fix' this back to True to match the other backends." With this change, when a use_kernel=True caller does not pass enable_telemetry, the key is omitted and the kernel applies its own default.

The resulting default is not verifiable in this repo — it lives in databricks_sql_kernel. If the kernel defaults telemetry_enabled to True (which would match the connector-wide enable_telemetry default of True at client.py:435), this silently flips kernel telemetry from opt-out to opt-in for users who never set the flag — a privacy-relevant default change that goes beyond the "let kernel own the default" framing.

Please confirm the kernel's built-in default matches the intended behavior here, and consider noting the effective default in the comment (or a test asserting the end-to-end default) so a future kernel-side change to that default can't silently alter connector behavior.