Add UAST to old FilePairs

[dpordomingo]

As described by https://github.com/src-d/backlog/issues/1242#issuecomment-372313046 ML team needs:

• [x] new 500 pairs added to existing experiment on staging,
• [x] to be able to add more annotations for the research,
• [x] add UAST to the already existent FilePairs

In order to accomplish that :point_up: we should: :point_down:

• obtain the new data from ML:
  • [x] add it into gdrive → https://github.com/src-d/backlog/issues/1242#issuecomment-372313046
• prepare (and pull request) the migration scripts:
  • [x] (A) to add 2 UAST cols (nullable) to the file_pairs table, → https://github.com/src-d/code-annotation/pull/221
  • [x] (B) to add to each FilePair from our file_pairs table, the UAST stored in the ML DB, → https://github.com/src-d/code-annotation/pull/221
  • [x] (C) (optional) to remove diff column since it is not longer needed (by https://github.com/src-d/code-annotation/pull/194), → https://github.com/src-d/code-annotation/pull/221
  • [x] (D) (optional) to add the 500 new FilePairs stored in the ML DB,
• proceed with the migration
  • [x] add into gdrive the current internal DB → infra#152,
  • [x] run the migration scripts locally over the downloaded DB, to obtain a new DB,
  • [x] add into gdrive the internal DB obtained from the prev step → infra#152,
  • [x] replace the current CAT internal DB, with the new one → infra#152,

Notes:

• The different versions of the internal DB will be shared in:
  gdrive/engineering/applicationsdemo/code-annotation-databases/3-add-UAST
• All the migration scripts could be done "at once".
• The (D) migration script was not needed because the 500 new FilePairs were added using the new Experiments Manager

[dpordomingo]

Issue desc update: The (C) migration script was not needed because the 500 new FilePairs were added using the new Experiments Manager

[dpordomingo]

done