Fix local YAML recipe scanning to be recursive

[kdvolder]

What's changed?

Fixes a bug introduced in #7338 where ClasspathScanningLoader failed to recursively scan subdirectories for YAML recipes when loading from a local path (e.g., target/classes during mvn test).

Replaced the non-recursive Files.newDirectoryStream with Files.walk in listYamlResourcesFromPath, and added handling for UncheckedIOException which can be thrown by the stream traversal if it encounters unreadable files/directories.

What's your motivation?

We have a large suite of YAML recipes organized into subdirectories (e.g., META-INF/rewrite/framework60/). After upgrading to 8.79.6, all of these recipes started failing to load during local Maven test runs with "Recipe(s) not found" errors.

Anything in particular you'd like reviewers to focus on?

The exception handling around the Files.walk stream. We catch both IOException (for the initial directory open) and UncheckedIOException (for any unreadable subdirectories encountered during traversal) to maintain the previous behavior of gracefully ignoring unreadable files.

Anyone you would like to review specifically?

@sam-snyder (since this relates to the changes in #7338)

Have you considered any alternatives or workarounds?

The only workaround for users currently is to flatten their entire META-INF/rewrite directory structure, which is not ideal for large recipe collections.

Any additional context

A witness test scanYamlInSubdirectory was added to ClasspathScanningLoaderTest to verify recursive YAML loading works correctly.

Checklist

• I've added unit tests to cover both positive and negative cases
• I've read and applied the recipe conventions and best practices
• I've used the IntelliJ IDEA auto-formatter on affected files

[timtebeek]

Thanks for the immediate fix!