fix: re-run macro argument validation after partial parse reuse (#12574)

[kalluripradeep]

Problem

Fixes #12574

When dbt parse is run a second time without any file changes, partial parsing determines nothing has changed (skip_parsing=True) and reuses the saved manifest as-is. In this path, MacroPatchParser.parse_patch is never called, so _check_patch_arguments is never triggered — causing macro argument mismatch warnings to silently disappear on all subsequent runs.

Root Cause

_check_patch_arguments was only reachable through parse_patch, which is skipped entirely during skip_parsing partial parse reuse. There was no mechanism to re-run validation against the cached manifest.

Fix

1. Extracted validation logic from MacroPatchParser._check_patch_arguments into a standalone module-level function validate_macro_arguments() in schemas.py, so it can be called from multiple call sites.

2. Added ManifestLoader._revalidate_macro_arguments() in manifest.py, which runs after self.manifest = self.saved_manifest when validate_macro_args is enabled. It re-extracts Jinja arguments directly from macro_sql and compares them against the YAML arguments stored in the schema file's dict_from_yaml, mirroring what MacroParser and MacroPatchParser do during a full parse.

Test

Added TestMacroArgValidationAfterPartialParse to tests/functional/macros/test_macro_annotations.py which:

• Runs dbt parse once (full parse) and asserts the warning fires
• Runs dbt parse again with no changes (partial parse, skip_parsing=True) and asserts the warning still fires

This is the exact repro case from the issue.

[github-actions]

Thank you for your pull request! We could not find a changelog entry for this change. For details on how to document a change, see the contributing guide.