cli tool. stream an exact time range from logs stored in s3. find the time segment with binary search so you download the minimum data possible.
s3-time-stream --bucket log-bucket --start 2017-04-27T16:08:00.212Z --end 2017-04-27T16:08:02.212Z
expects that logs are stored in s3 with date+hour prefix.
s3://bucket/2017-04-19T20
expects log lines to start with date in this format right now.
<134>2017-04-27T16:08:00Z
there can be many logs per hour they will be streamed in time order even if they overlap.
both of these things should be configurable later. e_notime