Added RepeatAttribute that is aware of RandomizedContext

[NightOwl888]

• You've read the Contributor Guide and Code of Conduct.
• You've included unit or integration tests for your change, where applicable.
• You've included inline docs for your change, where applicable.
• There's an open issue for the PR that you are making. If you'd like to propose a change, please open an issue to discuss the change or find an existing issue.

Summary of the changes (Less than 80 chars)

Fixes #1106

Description

This creates a new RepeatAttribute that is nested within LuceneTestCase. In addition to repeating the test X number of times, this attribute ensures that each iteration resets the test seed on RandomizedContext.RandomGenerator so failed test conditions can be replicated without the need to apply the RepeatAttribute.

This sets a precedent for how to test other NUnit extensions (such as attributes) so we can also add tests to confirm they behave as expected. Some bits of the NUnit test approach were added so we can integration test the RepeatAttribute to confirm the correct behavior. The commit cdfd62b includes additional asserts that verify:

• The same random seed is used for the duration of the test
• The same test seed is used for the duration of the test
• The test seed changes from one iteration to the next

Note that RepeatAttribute does not repeat the OneTimeSetUp() or OneTimeTearDown() during each iteration, so every repeated try will use the same culture, time zone, similarity, and codec. Perhaps we can do that at some future point, but it is more complicated to maintain the TestExectutionContext, RandomizedContext and seeds to make that happen.

The format of the random seed string has also changed to use 2 hexadecimal numbers separated by a :. However, existing seeds with one hexidecimal number can still be parsed to replicate test failures. This not only allows us to make features that update the seed, but also will make test replicability more stable across source code updates on tests.

[paulirwin]

This looks great, nicely done. The actual implementation is refreshingly simple! (Although I'm sure the work to get there was not.) I left one comment that could be looked into, but not a must-have change.