Compress PDF

It depends entirely on what is inside. A file whose streams were saved loosely gives up a few percent; one already packed tightly might not shrink at all, and a scan whose images are JPEG already comes back the same size, because the default pass never touches image data. The tool runs qpdf with object streams and Flate recompression at its highest level, then reports the size before, the size after and the percentage.

Not by default. The standard pass only repacks the streams and structure of the file, so no page is rasterised and no image detail is thrown away. Turning on image re-encoding does trade quality for size: it rewrites embedded images as JPEG, which shrinks a photo-heavy file a great deal and softens it visibly. That is why it is off unless you ask for it.

Because compression lost. Object streams and a cross-reference stream carry fixed overhead that, on a small PDF, can cost more than the recompression saves. Rather than hand back something larger, the tool keeps your original bytes and says so: "already optimal, returned unchanged".

One, and it is a switch rather than a slider: image re-encoding, off by default. Everything else is fixed at the strongest lossless setting qpdf has, so there is nothing left to trade. A dial claiming finer control over a pass that keeps every byte of image data would be measuring nothing.

No, and it says so rather than failing quietly: "password-protected, unlock it first". Run Unlock PDF with the password, compress the result, then protect it again if you need to.

Yes. Drop as many as you like, each is compressed separately and comes back as its own file, named with a -compressed suffix.