Various improvements to integration

[GeigerJ2]

Without data connector workaround

• Fetch metadata from MCA via Python script (DOI and file names of entries; dump to JSON), and replace again jinja-style variables in startup notebook
• "Cache" archive_url and if second session is started and env variable is different from expected one, notify the user to shut down the first session
• Show list of files in the MCA archive entry and let user select (in appmode) which one to use, and do the wget there, not on startup

Appmode

• Have everything in appmode, three boxes, 1) work in jupyter mode, 2) show statistics, 3) ...

Misc

• How much storage at max -> 64 gb
• Copying over a 1 gb archive over from data connector mount took 10s, while cp on same file system took twice as long -> Network is not limiting factor, faster than disk I/O
• We had problems with verdi config set without a profile, which would except, but now it works.
• Debug slow profile creation with sqlite_zip on 0.5 cpu and 10gb mem machine on Renku
• They seem to use rclone
• Mounting ourselves using that, probably not possible
• Use different template, if not created from MCA and archive_url set

Make archive download and profile creation cells hidden using

{
  "jupyter": {
    "source_hidden": true
  }
}