Introduce data generators as virtual test cases

[Garados007]

Each data generator is now wrapped as a virtual test case which is also reported into the test results. To achieve this the setup for data generator has been changed into three phases:

1. Loading the list of data generate and do some general checks (for example if the data generator can be found or its function signature is correct).
2. Execute all data generators one by one and wrap each of them in their own test case. Do some basics checks with the generated results (for example if an invalid wave is returned).
3. Assign these data generators to their dependend test cases. Do some checks (for example if the types of the generated results match the function signature of the test case).

Close #353

---

This PR does also consists of two refactoring steps. In one refactoring commit are all data generator, multi data and multi-multi data test related functions moved in their own file which reduces the size and complexity of UTF_Basics a lot. The second refactoring commit moves some polyfill methods into their own file.

[Garados007]

@t-b @MichaelHuth Ready for review.

[Garados007]

@MichaelHuth I have changed it in that way that it will call ReportErrorAndAbort if I get an invalid wave back from the data generator. This will report the error and stops executing RunTest without any writing JUnit or TAP output. This should be the same behavior as before if I am right.

The other idea with collecting the errors and printing JUnit only for the failed data generator suite is quite interesting but has some problems:

• I have to somehow mark the run abort. I can abuse setAbortFlag() for this but I have to catch it at some point and I don't know if it will have some weird side effects.
  • Another idea is to check if there are any assertion errors but there is no difference with CHECK errors.
• For printing the JUnit log I have to call the TEST_BEGIN and TEST_END hook as its also writing TAP. And I don't know if its good to call both hooks (as they also execute user hooks) when the data generator failed.
• Phase 3 of data generator setup can trigger some errors (e.g. invalid wave types) which cannot be assigned to a virtual test case right now.

[Garados007]

@MichaelHuth Please re-review this.

[t-b]

Assigning @MichaelHuth as I think this was the intention.

[Garados007]

@MichaelHuth Ready for review.

[t-b]

@MichaelHuth Just to be sure that this works now: Did you try to run one MIES hardware test with this PR?

[MichaelHuth]

I did not wrote that it works now. I was about to do some MIES tests as final test.

@Garados007 It turns out that the setup for test cases for the data generators is not complete as I get an IUTF internal RTE when using an assertion in a data generator function from a non-existing igor_debug_assertion in DebugFailedAssertion.

If the global is created the MIES HW tests start running...

[t-b]

I did not wrote that it works now. I was about to do some MIES tests as final test.

Thanks, sorry for being impatient.

[Garados007]

I have refactored the debug helper methods into their own file. This allowed me to copy the set/reset of the Igor Debugger from regular test cases to the data generators. During this fixed also the bug with the non-existent igor_debug_assertion as it is always created before each data generator execution.

[Garados007]

@MichaelHuth Ready for review.

[t-b]

Thanks both!