:sparkles: Features
* Add flag for preserving unit test fixtures to improve debuggability (2404) georgesittas
* Support variables in native SQLMesh projects (2391) izeigerman
* Improve lineage perf (2316) tobymao
* Support restatements for the incremental unmanaged model kind (2423) izeigerman
:beetle: Bug fixes
* Randomize unit test fixture names to avoid concurrency issues (2403) georgesittas
* Make a 1-1 correspondence between created/dropped fixtures in unit tests (2407) georgesittas
* The message used for prompting the backfill end date in CLI (2408) izeigerman
* Make partial setting in unit tests deterministic (2410) georgesittas
* Widen spark support (2409) eakmanrq
* Make Snowflake table part parsing more lenient (2412) georgesittas
* Avoid forcing quoted identifiers when creating views in trino and spark (2415) izeigerman
* Improve csv seed file type inference (2418) georgesittas
* Get physical table name in ibis example (2417) andrew-sha
* Ensure lineage doesn't mutate the model's query tobymao
* Use end-bounded missing intervals when checking for gaps during the promotion (2424) izeigerman
:hammer_and_wrench: Refactor
* Only use the last comment attached to a projection as its description (2421) georgesittas
:broom: Chore
* Use default_dialect property instead of config.dialect in context (2405) georgesittas
* Update databricks doc to link to notebook magic (2402) eakmanrq
* Only print the environment name in logs during the environment finalization izeigerman
* Chore! bump sqlglot to v23.8.0 (2411) georgesittas
* Fix typo in plans.md (2422) VaggelisD
* Add error message for untyped self references (2419) tobymao
* Document expected column order in unit tests (2427) georgesittas
:alien: Other
* Update mssql test from bytes to timestamp return value (2406) treysp
* Wordsmith comment registration docs (2416) treysp
* Clarify allowed cron values (2425) treysp
* Update duckdb incremental by unique key materialization strategy (2426) treysp