Make autotests accurately represent which tests are valid for each platform

Currently, the metrics system reports too many test as being valid for some platforms. One of the reasons for this is that there are some test programs and a lot of test functions that are effectively empty for certain platforms because they have most/all of their content ifdef'd out.

The existing autotests must be reviewed and corrected to ensure that:
1. Test programs that are not useful on a particular platform are neither built nor executed on that platform.
2. Individual test functions that need to be excluded on a particular platform are excluded by ifdef'ing out the entire function instead of ifdef'ing just the function body and leaving an empty function to inflate the pass-rate.

Additionally, the correct way to exclude test programs and test functions should be documented in the qtestlib manual.