Mesos fetcher cache doesn't retry when missed.

Mesos Fetcher doesn't retry when a cache is missed. It needs to have the ability to pull from source when it fails.

421 15:52:53.022902 32751 fetcher.cpp:498] Fetcher Info: {"cache_directory":"\/tmp\/mesos\/fetch\/slaves\/<slaveid>","items":[\{"action":"RETRIEVE_FROM_CACHE","cache_filename":"<file.name.tar.gz>)","uri":\{"cache":true,"executable":false,"extract":true,"value":"https:\/\/<file.path.s3.amazonaws.com>\/<file.name.tar.gz>"}}],"sandbox_directory":"\/var\/lib\/mesos\/slave\/slaves\/<slaveid>\/frameworks\<frameworkid>\/executors\/name\/runs\/<id>"}
I0421 15:52:53.024926 32751 fetcher.cpp:409] Fetching URI '"https:\/\/<file.path.s3.amazonaws.com>\/<file.name.tar.gz>"
I0421 15:52:53.024942 32751 fetcher.cpp:306] Fetching from cache
I0421 15:52:53.024958 32751 fetcher.cpp:84] Extracting with command: tar -C "\/var\/lib\/mesos\/slave\/slaves\/<slaveid>\/frameworks\<frameworkid>\/executors\/name\/runs\/<id>' -xf '/tmp/mesos/fetch/slaves/f3feeab8-a2fe-4ac1-afeb-ec7bd4ce7b0d-S29/c1-docker-hub.tar.gz'
tar: /"https:\/\/<file.path.s3.amazonaws.com>\/<file.name.tar.gz>": Cannot open: No such file or directory
tar: Error is not recoverable: exiting now
Failed to fetch '"https:\/\/<file.path.s3.amazonaws.com>\/<file.name.tar.gz>"': Failed to extract: command tar -C '"\/var\/lib\/mesos\/slave\/slaves\/<slaveid>\/frameworks\<frameworkid>\/executors\/name\/runs\/<id>' -xf '/tmp/mesos/fetch/slaves/<file.name.tar.gz>"' exited with status: 512