Tango should skip unavailable backends when using distDocker

[20wildmanj]

From Chaskiel re: an error that recently occured on CMU prod

The error was triggered by an operating system error (an LDAP lookup failed when tango ssh'd to the docker node). There has only been one such failure since the beginning of the semester

ERROR|2024-02-08 10:43:41,556|TangoREST|addJob request failed: Command '['ssh', '-o', 'BatchMode=yes', '-i', '/usr/local/lib/Tango/vmms/id_rsa', '-o', 'StrictHostKeyChecking=no', '-o', 'GSSAPIAuthentication=no', '[email protected]', '(docker images)']' returned non-zero exit status 255.

For the devs: the failure was in getImages, not waitVM (which has retry logic), or a later step (where retries are more complicated because you'd have to restart the process from the copyIn phase) It may make sense for DistDocker.getImages (and maybe DistDocker.getVms) to skip backends that are unavailable.