The original intention was to change the watchpoint Python test to a lit test. While doing so, I found a situation where stop-command-source-on-error was being ignored. Since the new test also tests the fix for the CommandObjectCommands, I figured I'd keep them in a single patch.
- I converted the test to lit because it is flaky on GreenDragon.
- The problem with the command object is that we didn't differentiate between the boolean value not being set, and the value being set to either true or false. If the value isn't set, we'd calculate it in the command interpreter based on the global options. With the old approach, we'd always set the value to the default of the option, which is not what the user wanted.