Thanks for the mention, Doitsu. Since my tools came up, here is where they stand, in case they help with what you are testing.
Most of my time right now goes into reading the EPUB specifications and turning them into test books. Each book in
epub-test-books checks one rule, and I write the expected result from the specification before any validator runs on it. That has shown me bugs in
epubveri, my own validator, that a shelf of real books never would have, and I fix them as they come up. It has also turned up places where epubcheck disagrees with the specification, and I report those on
its issue tracker: 19 issues so far, five of them closed.
epubsana is the repair side. It takes the defects epubveri reports, proposes the ones that have exactly one correct fix, asks before each change, and writes the result to a new file, so the original is never touched. Some of what the test books uncover is now repaired automatically this way. What needs an editor's judgement, it leaves alone.
My time is limited, so the test books and epubveri come first. With what is left I fix bugs in epubsana,
kepubverto (EPUB to Kobo KEPUB),
imagoverto (image conversion),
paginaverto (PDF and page images to a structured document) and
epublift.
All of them are 100% free software, with their full source code public under the AGPL, and I will keep developing them in the open as my work and private life leave time. One of my goals is to build them in pure Rust, with no C libraries and no Java runtime. The test books themselves are CC BY 4.0, so anyone can copy them into their own test suite. If you want to help, try them on your own files and report what goes wrong. A book that breaks one of my tools is the most useful bug report I can get.