Fix hunk apply failure caused by trailing whitespace mismatch

[bmuschko]

Summary

Fixes PatchApplyException when applying multi-hunk patches where the source file has trailing whitespace on context lines that the patch does not (or vice versa).

Before: Context lines in canApplyAt() were compared using strict ByteBuffer.equals():

|| !newLines.get(pos).equals(slice(hunkLine, 1)))

This required an exact byte-for-byte match, so "Line 118 " (with trailing spaces) in the file would not match "Line 118" in the patch context, causing the hunk to be rejected with:

PatchApplyException: Error applying patch in test.txt, hunk HunkHeader[118,6->124,7]: hunk does not apply to file content
    at org.openrewrite.jgit.api.ApplyCommand.applyText(ApplyCommand.java:676)

After: Context lines are compared using equalsIgnoringTrailingWhitespace():

|| !equalsIgnoringTrailingWhitespace(newLines.get(pos), slice(hunkLine, 1)))

This strips trailing spaces, tabs, and \r from both the file line and the hunk context line before comparing. Since context lines are only used for positioning the hunk (they're not modified), requiring an exact trailing-whitespace match is unnecessarily strict and causes valid patches to be rejected.

Additionally fixes FileHeader.parseName() to strip trailing \r from parsed file paths when patches have CRLF line endings.

[pstreef]

Looks good. The core change in canApplyAt() is well-motivated — context lines are only used for positioning, so requiring exact trailing-whitespace matches is unnecessarily strict.

Verified that RawText.getRawString() strips \n but leaves \r from CRLF lines, which is exactly what equalsIgnoringTrailingWhitespace() handles. The FileHeader.parseName() fix for CRLF patch paths is clean too.

Minor nits (non-blocking):

• Tests use JUnit assertTrue — AssertJ (assertThat(...).contains(...)) would give better failure messages
• A negative test for a genuinely mismatched context line would add confidence the comparison still rejects real mismatches

[pstreef]

oops duplicate