Remove workaround for ghost jobs caused by stale reads from SDB

[cket]

~~AWS is moving away from SDB in favor of DynamoDB. Switching would solve the inconsistency problems seen in #760 and increase scalability.~~

Once we moved the AWSJobSTore from SDB to DynamoDB, we should remove the workaround for #760. [edited by @hannes_ucsc]

┆Issue is synchronized with this Jira Story ┆Issue Number: TOIL-96

[hannes-ucsc]

The way its written now, it would be a dupe of #964, but I will rewrite it.

[adamnovak]

This also seems to need to apply to stale reads from inconsistent filesystems.

[adamnovak]

This is going to be fixed by the fix for #3841.