Improve file format detection

For now the library detects the format of a file (JPEG, PNG, ...) based on its extension. What if a file has no extension? The file could even have the wrong extension, for example some CDNs cache and deliver in the WebP format files that are originally JPEG or PNG.

The library should detect the format of a file by looking at its content instead.

Start by locating the library's current extension-based format detection. Review how JPEG, PNG, and WebP content can be distinguished, then define coverage for files with missing or incorrect extensions. Done means the library identifies the format from file content rather than relying on the filename extension.

Written by the indexing model from the issue text.