move chunk metadata from seek to data file name

[alfred-landrum]

With the recent changes in archive.Import, when we create a new data file, we'll know the first & last record timestamps, as well as the record count, for the data file. So we no longer need to encode that information in the seek file name, which will let us simplify the walk code that parses file names & identifies chunks.