Introduce basic benchmarks to UTF

[Garados007]

After discussing #374 we came to the conclusion that we want to implement the following features in the first step:

• Add a function tag UTF_BENCHMARK
• If test is marked with this function tag it doesn't allow assertions at all. This is done by checking the assertion counter after the test case if its equal to zero (normally, it should be different to zero).
• The function tag require an integer value which represents the number of cycles.
• The test is run repeatedly according the number of cycles. If one iteration failed the benchmark is aborted.

[Garados007]

There is a proposal for the timing measurement. I don't remember what our decision was on this topic.

• The benchmark is executed by repeatedly calling the test function directly. ExecuteHooks with TEST_CASE_BEGIN and _END is only called at the begin and end of the benchmark and not somewhere in between. This also results in one test case in the results and one time measurement for all executions.
  • Doing this we have as few overhead for the benchmark run as possible and we can get useful data without polluting the test results with 10,000 iteration. Also isn't the data for each step useful as we are interested for the results of the whole benchmark run.
• The timing measurement for the end time is corrected by this formula: end_time = start_time + (end_time - start_time) / cycles. With this we have the average execution time in the test results without knowing the number of cycles at a later point of time.

[t-b]

I don't remember what our decision was on this topic.

IIRC we wanted to add a function tag to define the number of cycles.