Since you asked and it was recommended to me some time back to chat with you about this... ;)

The state of audio restoration tools on Linux is abysmal.  The short version is, nothing on Linux comes close to options in other OSes.  There are some options to run via Wine (ie. I run WaveARTs plugins via Wine) but many more options that do not (ie. iZotope).

The tools I can think of...

Hum/Buzz Removal -- Essentially Notch filters, that are harmonically linked.

Broadband Noise Removal -- Even a good multiband gate can help, as that is close to what WaveARTs MRNoise is, and it is noticeably better than options on Linux.  I am not sure what iZotope does for it's destructive process, but I believe is also a form of multiband gating.

Click/Pop removal -- Not something I use a lot, so can't comment here

Expander -- A decent expander is surprisingly difficult to come by.  Most options on Linux are gates instead of true expanders, and the expanders I can find like Calf, I haven't been to happy with the results I have been able to get out of it.  I would suggest looking at MRNoise from WaveARTs for a great example of this, it honestly is one of the best expanders I have used.

There is one last component to this, which is a destructive environment to work in, actually non-realtime is a better term for what I am looking for.  Believe it or not this is one of the few areas I personally believe a non-realtime workflow is better than realtime/non-destructive.  I prefer being able to find areas with a spectrograph to sample to set my multiband gating process off of, and then apply to areas in the same way.  I prefer to do this non-realtime because in the same audio clip delivered to me by a video editor, I may need to do this multiple times, with different settings each time, for each edit point they have at times.

The workflow I would ideally see is the ability to open the file in the editor, apply Hi/Lo-pass filtering as appropriate, hum and buzz removal if needed, sample a section of audio for setting any appropriate settings for hum/buzz and for broadband noise removal, then finally remove the noise via multiband gating or whatever.  The last step for me personally is downwards expansion to hopefully provide just a slight amount more cleanup at the end.

The use case is pretty much a given for anyone that gets video from editors in any form(But especially dialog), and has to edit the audio together.  It also works wonders if used appropriately on field recordings for SFX, Dialog, etc.   It is the one thing I wasn't able to stay in Linux for when working on the recent Tube trailer.

So that should be a few years worth of work for you, should you choose to accept it:)

         Seablade