Speed up getFiles by start/stop time (Closes #23)

[jtniehof]

This PR speeds up the calculation of possible output files and thus closes #23. This works by JOINing the file table with a table containing Unix times and doing the search using that table. If that table doesn't exist, the old (slower) code will be used. Included is a script to migrate databases by adding this table and populating it with existing file entries; eventually this script can be expanded to perform other migration tasks (e.g. the yesterday/tomorrow columns) as part of #7.

PR Checklist

• Pull request has descriptive title
• Pull request gives overview of changes
• New code has inline comments where necessary
• Any new modules, functions or classes have docstrings consistent with dbprocessing style
• Major new functionality has appropriate Sphinx documentation
• (N/A, see below) Added an entry to CHANGELOG if fixing a major bug or providing a major new feature
• New features and bug fixes should have unit tests
• Relevant issues are linked in the description (e.g. See issue # or Closes #)

No changelog entry since it doesn't exist, release notes pending (#28)

[jtniehof]

@dnadeau-lanl, @balarsen has requested that you do the review as part of his stepping back from the project, so I've changed the request.

[jtniehof]

Plan is to roll this out to PSP/ISOIS product as part of the current downlink. I've tested it in isolation as much as possible and I feel pretty good about it, and this'll give it a good beating in production. I think it can still be merged to master; doesn't need to wait for further testing.

Science data for this downlink should be completed by Friday 16 October. I'll spend some time next Monday reviewing the logs and results just in case anything crops up, and assuming that's good it can deploy to ECT. If anything does come up, I'll get in a unit test to catch it and either add to this PR or open another one.