Update coverage file in BabelStream case study

We forgot to update the BabelStream case study when we changed the coverage
schema. Running the example as it was gave an error due to schema validation.

This commit is a best-effort attempt to convert the old coverage file into the
new format. It is "best-effort" because:

- The old format did not include filenames, but the new format requires them.
  Since the case study already uses anonymized platforms and languages, the
  simplest solution here was to generate names like "file1" and "file2".

- The old format stored lines as regions containing the number of "real" lines
  of code in the range (start, end), rather than storing the specific lines of
  code that were used. However, since all the data was generated by the
  same version of Code Base Investigator parsing the same files, all the
  regions line up. The simplest solution therefore was to pretend that the
  region represented a contiguous run of lines. The information about how the
  number of comments and amount of whitespace in a region is destroyed by this
  transformation, but it isn't necessary to compute divergence.

Signed-off-by: John Pennycook <[email protected]>