Run page improvements

[ewels]

In https://github.com/MultiQC/website/pull/66 I added a new page where you can run MultiQC in the browser, using WebAssembly.

It works ok, but there are some improvements that could be made:

• [x] File listing in the browser is a bit basic, looks a bit rubbish
• [x] Rewrite to use Pyodide web workers running on separate threads, instead of locking up the main browser thread.
• [x] Replace file mounting method
  • Can only mount a single directory, once and can't mount files, only directories
  • showDirectoryPicker method is experimental and doesn't work in many browsers
  • Could maybe use something like this? (demo)
• [ ] Button to zip + download multiqc_data directory
• [x] Allow new report without page reload
  • If more files dropped on file dropper after report generation, clear previous report and continue.
• [ ] [Stretch] Have a accordion / modal with a form to set some config options, eg. title / theme etc.

[ewels]

I think for the filesystem we need to ditch the File System Access API completely and go back to something a bit more regular like this to scoop up files from the browser UI. Can hopefully tie those listed files into a pyodide filesystem interface such as MEMFS. Could allow repeated drops to continually add files then and keep updating the list from the Python side.

[ewels]

@vladsavelyev figured much of this out, now live on seqera.io/multiqc 🎉