Builds should fail when there is a problem with test processing

[radeklawgmin]

From time to time there is a problem with test processing. It results in having the status as in case when there are some changes spotted. In such cases there should be a feature allowing the user to switch how the build should behave in such cases. What about having two cases:

• When there is a processing error the build should fail (to distinguish from builds mark in red when there are some changes spotted),
• When there is a processing error the build should behave as it does now (the build should not fail and the report can be open in the browser although there are some or no results). The actual behaviour of the build should be configurable by the user.

[tMaxx]

See also #80, which elaborates on this a bit. These two tickets ideally should be implemented at the same time.