docs: clarify S3 builder URL path handling

[ZeroAdark]

Which issue does this PR close?

Closes #409.

Rationale for this change

The AmazonS3Builder::with_url documentation used a path-bearing URL without explaining that the builder applies connection information but does not retain the path as a store prefix. This could lead callers to read from the bucket root unexpectedly.

What changes are included in this PR?

• Clarify that URL paths are not retained as prefixes.
• Point callers to crate::parse_url when they need both the store and path.
• Use a bucket-only URL in the builder example.

Are there any user-facing changes?

Documentation only; there is no API or runtime behavior change.

Validation

• cargo fmt --all --check
• targeted AmazonS3Builder::with_url doctest
• cargo test --doc --all-features (54 passed, 2 ignored)