The populate_sdk_test is used for sanity test after the sdk is built from the DEV side, we usually build the SDK, but don't know whether it works or not from the DEV side, when the QA team reports the problem, then we need pay a few time to reproduce and fix it, we can add a task like populate_sdk_test (or other names) to test the sdk after it is generated, and find the problem as early as possible rather than delay it to the QA team.

The basic design is:
1) addtask populate_sdk_test after populate_sdk
2) The populate_sdk_test will install the sdk and run some test cases when SDKMACHINE is the same as the build host, if not, then skip the test and warning.
3) We need the ability to add the testcases to populate_sdk_test