HHH-19472: native queries can return Object[]

[victornoel]

The fix is quite simple and obvious. I also refactored the test that was previously introduced for HHH-18450 to cover more cases and make it easier to maintain.

My understanding is that when executing queries (any kind), we either provide a row transformer, a tuple transformer or the expected result class. Sqm-based queries are providing the row transformer directly (based on the result class) while native queries are currently implemented with providing directly the tuple transformer.

So I went with fitting in the existing code by supporting Object[] at the tuple transformer level for native queries.

Alternatively I looked into producing a row transformer in NativeSelectQueryPlanImpl or moving all the default row transformer instantiation (that can be found in a few places of the codebase, e.g., ConcreteSqmSelectQueryPlan#determineRowTransformer) directly inside JdbcSelectExecutorStandardImpl to reduce code duplication and align behaviours but it was way too much work and I worried to introduce regressions.

See also #10177 for 6.6 branch.

---

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license
and can be relicensed under the terms of the LGPL v2.1 license in the future at the maintainers' discretion.
For more information on licensing, please check here.

---

https://hibernate.atlassian.net/browse/HHH-19472

[mbladel]

LGTM, once @victornoel fixes CI we can merge this.

[victornoel]

Ok so as recommend by @mbladel in #10177, I added @RequiresDialect(H2Dialect.class) as well as for PostgreSQL (since it worked and it's good to at least test this on 2 dbs) because some db does not even accept select 1 while some other returns Long instead of Integer in some cases… I hope this time it's mergeable :)

[gavinking]

I think we can merge this now.