Configure Spark client timeouts using Hadoop time durations

[arielshaqed]

Instead of hand-crafted parsing routines, use Hadoop Configuration.getTimeDuration for our timeouts.

[github-actions[bot]]

This issue is now marked as stale after 90 days of inactivity, and will be closed soon. To keep it, mark it with the "no stale" label.

[arielshaqed]

@ozkatz probably useful before a possible "1.0" release. (This is Spark metadata client, mind. lakeFSFS is different because it lives in the Hadoop ecosystem where raw numbers rule...)