Fix download using spine information from window.bData

[ahxxm]

This is similar to my another PR for the new format, the branch is based on #59

audiobooks work just fine with command odmpy libby --ebooks --direct --chapters --merge

for epub books command is the same, and I picked a small one to test locally,

• it can generate an EPUB correctly with ToC, text styles like bold and italic are preserved(since they are from base64-encoded HTML?)
• images are also saved, but they didn't exist in bData anymore, I'm not 100% satisfied with the way they are downloaded. e.g. when image src is ../Images/cover.jpg, it can still be downloaded, but need a fix on HTML to display, instead of preserving the origin file hierarchy
• CSS is downloaded similarly

There are 2 commits trying to make CI happy, but apprently CI isn't that easy to please: linter passed, tests hardcoded the outdated openbook.json and should be fixed as well

[ahxxm]

@ping I spent some time to understand mocks but did't make meaningful progress, it would be great if you can help

[bckelly1]

I was trying to figure out why this doesn't work. I think the spine content changed: "spine": [{"path":"_d/record_access_violation"}] Looks like they changed the API again.

[ahxxm]

I was trying to figure out why this doesn't work. I think the spine content changed: "spine": [{"path":"_d/record_access_violation"}] Looks like they changed the API again.

yeah, it's bad, please don't use this branch now, the path looks surveillant