[lennartkats-db]

Summary

Preserve literal trailing and repeated underscores from a custom presets.name_prefix when naming bundle schemas. Keep the prefix generated by development mode on its existing normalization path.

Before

Issue #2305 reports that a schema named qc in my_catalog becomes my_catalog.my_userqc when the custom prefix resolves to my_user_. The issue expects my_catalog.my_user_qc and also reports that repeated underscores are collapsed.

After

ApplyPresets now preserves the literal underscores in custom schema prefixes while normalizing the other prefix segments. For the reported example, the schema name is my_user_qc. The change is limited to schema names; the generated development prefix retains its existing behavior.

Verification

• The host's focused Go regression check, TestApplyPresetsSchemaPrefixPreservesUnderscores, failed at base 3aa5c9607bff73d3fb18a4c6e63b9f03235f3de7 and passed at head 6692d77a3f251a4c32024aa3007d41a5ebf1e6a3. The cases cover trailing and repeated underscores, a bracketed custom prefix, and the generated development prefix.
• A remote databricks bundle deploy was not exercised in the saved test receipt. Current-head CI remains a publication gate.

Review notes

Existing deployments using a custom prefix with literal underscores may receive a different schema name after this change. The host records the independent output review as skipped under the pilot policy; this is not an independent approval.

[eng-dev-ecosystem-bot]

Integration test report

Commit: f04aa2d

Run: 41506606023

	Env	✅​pass	🙈​skip	Time
✅​	aws linux-2core-8gb	281	15	7:42
✅​	aws-windows-latest-4core-16gb	283	13	4:16
✅​	azure linux-2core-8gb	280	15	6:53
✅​	azure-windows-latest-4core-16gb	282	13	3:25
✅​	gcp linux-2core-8gb	281	15	8:10
✅​	gcp-windows-latest-4core-16gb	283	13	4:24

Top 3 slowest tests (at least 2 minutes):

duration	env	testname
3:50	azure linux-2core-8gb	TestAccept
3:46	aws linux-2core-8gb	TestAccept
3:43	gcp linux-2core-8gb	TestAccept