getGlobal returned NULL error after upgrade

Using OCR capability provided via Intelligent Quality add-on to test a LabView application. TestComplete python scripts were working fine on TestComplete ver. 15.69. Upgraded to ver. 15.73 and am now intermittently hitting the following error:

Run test script 1 - this automatically invokes application and script is run successfully. Application is left running.
Run test script 2 - script begins but then immediately hits "Python runtime error. SystemError: <built-in function getGlobal> returned NULL without setting an exception. The error occurs on a simple variable assignment (e.g., "map_sel_pref": spNone) within a python dictionary. The problem appears to be in identifying the spNone (which is one of the SelectionPreference constants associated with the OCR BlockByText() method). 

If I shut down the application and then re-run test script 2, the test script will invoke the application and test script 2 will be successful.

This does not happen consistently -- there are cases where steps 1 and 2 above run fine without error. In cases where I hit the error, I can successfully access the NameMapping objects -- that is, TestComplete can identify the application after test script 1 completes and the application is left running. 

Since the SelectionPreference parameter is simply an integer, I could change my code from "map_sel_pref": spNone to something like "map_sel_pref": -1 if I knew the constant values, but I don't believe that is documented anywhere. More importantly, not sure why this has become a problem with the latest update.

Any suggestions on fixing/resolving this?