IDA plugin: fix result save/load for same-named binaries

[xeropresence]

Problem

Diffing two databases with the same basename (e.g. v1/app.exe.i64 vs v2/app.exe.i64) produces a broken result
file:

• WriteResults copies both .BinExport sidecars next to the .BinDiff by basename; the secondary overwrites the
  primary.
• DatabaseReader::Read resolves <dir>/<file.filename>.BinExport per side; with identical filename columns, the
  secondary export is loaded as both call graphs.

Loading such a result then shows the "original file hash … is different from the primary one in the result file"
warning, and all primary addresses are resolved against the wrong call graph.

Fix

• ida/main_plugin.cc (ExportIdbs): when both IDB stems are equal, export as <stem>_primary.BinExport /
  <stem>_secondary.BinExport — the same convention the Java UI uses
  (NewDiffImplementation.createUniqueExportFileName). This flows through CallGraph::GetFilename() into the
  file.filename column, so save, load, and the UI all resolve distinct sidecars. Non-colliding names are unchanged.
• database_writer.cc (DatabaseReader::Read): fail with a FailedPrecondition if primary and secondary would
  resolve to the same .BinExport, instead of silently loading one file twice.

Notes

Result files already saved with colliding names cannot be repaired (the primary sidecar is gone); re-diff and save
again.

[google-cla]

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.