Introduce RetryAttribute for test methods

[Youssef1313]

Fixes #3161

[nohwnd]

FYI From previous talks with Playwright where they have good retries, that imho were attempted to be duplicated in mstest.playwright, Max also mentioned that they wanted:

• max retries to be configurable via config
• see what max retries value is in assembly initialize (might be specific to them because they are in position of test framework in vstest, so that story might not be relevant here, where they have way more power via extensions easily)
• have good reporting of the retries, to be able to show retries that were not passing

[Youssef1313]

max retries to be configurable via config

Not sure what shape would you be expecting. Currently, every application of the attribute is specifying the number of retries

---

see what max retries value is in assembly initialize (might be specific to them because they are in position of test framework in vstest, so that story might not be relevant here, where they have way more power via extensions easily)

This is currently info specific to the test method itself and is irrelevant to assembly initialize. I think you are talking of a completely different design that allows a "global" retry count, not a Retry attribute on test methods which I'm implementing here.

---

have good reporting of the retries, to be able to show retries that were not passing

We are going to have a design that allows that, but that is unlikely to be implemented in this PR though, unfortunately. We need to know better how to communicate that information to tools that will make use of it.

[Evangelink]

Shall we use an interface instead to give us more freedom in future without breaking change? I don't know if that's support useful or not given roslyn mainly goes with struct.