[hma] Allow /status for index staleness threshold to be configurable

[Dcallies]

With very large indices, it can become too slow to download the index than the hardcoded limit of 65 seconds (which means in reality you only have 35 seconds to finish downloading a new index).

Very large indices will eventually hit this, so we should make the liveness check (defense against the index refresh breaking entirely) configurable by OMM config.

Additionally, we should log how long it takes to refresh the config in the current index task.