[S3] Fix S3 list_recursive filtering with paths containing consecutive slashes

[npow]

BEFORE commit 21f29e2 (no filtering logic):

1. S3ROOT = "s3://bucket/path/" (with trailing slash)
2. User does: S3(s3root=f"{S3ROOT}/{test_prefix}/log/")
3. This creates: "s3://bucket/path//test_prefix/log/"
4. S3.init does rstrip("/") → "s3://bucket/path//test_prefix/log" (double slash still there!)
5. When listing, S3 API is queried with prefix path//test_prefix/log
6. S3 DOES return results because S3 ignores the double slash in the middle! S3 treats path//test_prefix same as path/test_prefix
7. All keys returned by S3 are included without filtering
8. Result: Everything worked fine!

AFTER commit 21f29e2 (with filtering logic):

1. Same setup, S3 API returns results (S3 still treats // as /)
2. BUT now the filtering logic kicks in:
   - prefix_url.path = "path//test_prefix/log"
   - normalized_prefix = "path//test_prefix/log/" (just adds trailing slash)
   - Checks if key_path (e.g., "path/test_prefix/log/test.txt") starts with "path//test_prefix/log/"
   - NO MATCH because path/ ≠ path//
3. All results are filtered out!
4. Result: Tests fail with empty results

[nflx-mf-bot]

Netflix internal testing[1390] @ 2280207