fix: Support OCI and compressed layer extraction

[Vortelf]

Fixes #687

Problem

• The Extract function was failing to extract files from Docker image layers for OCI-format images (the default since Docker 25+)

Root Causes

1. Layer ID extraction was wrong for OCI format images

• dive/image/docker/layer.go used strings.Split(tree.Name, "/")[0] to derive the layer ID which only works for Docker images and therefor it was always failing to retrieve the layer for OCI images.
• ExtractFromImage searched for entries named <layer-id>/layer.tar and OCI images store layers as blobs/sha256/<digest> which was also causing a mismatch when searching the source from which to extract the files.

2. No compression detection for OCI blob layers

• OCI layers are stored compressed (gzip/zstd), while Docker layers use plain tar files without compression, so simply finding the correct layers wouldn't work on it's own.
• zstd.NewReader constructor validates input, returning nil for non-zstd data and only failing later with magic number mismatch during reads
  • This piece is worth double-checking (func extractCompressedLayer(reader io.Reader, target string) error) https://github.com/wagoodman/dive/compare/main...Vortelf:dive:fix-file-extract-oci?expand=1#diff-41e690917a0067b8cdff96efaf54c45d60f9a80e17b2f12bc622aaa2086e2f94R349

Misc

• Tidied up the error handling a bit around this functionality, because it was constantly failing silently.