[evlatests] Scan boundaries continued

[Michael Rupen]

Martin put in a fix for the first-integration problem; we tested this
using the same scripts and it works!  So now all the integrations seem
to be assigned correctly to the appropriate scans.

Ken also did a quick test, integrating 3 LTA dumps in the CBE before
writing the binary data.  Again this worked perfectly.

There are a few remaining edge cases to check, but so far everything
now looks quite proper.

                  Michael