Allow ConditionalTheory to have an empty data collection about testing

[Tratcher]

SkipIfNoData?

[Eilon]

Adding a SkipIfNoData param sounds good to me.

[cesarblum]

This is messier than I thought it would be. The problem is xunit really doesn't like empty data:

https://github.com/xunit/xunit/blob/1b69df1378cb9d6b589d55ce23ab544458169485/src/xunit.execution/Sdk/Frameworks/TheoryDiscoverer.cs#L242

We'd either have to write our own discoverer that's basically a copy of xunit's without that check, or do something ugly like:

if (results.First() is ExecutionErrorTestCase &&
    (results.First() as ExecutionErrorTestCase).ErrorMessage.StartsWith("No data found for"))
{
    ...
}

Given we worked around our issue in Kestrel by introducing a new attribute that checks data will be available for the test, I'm thinking of close this issue.

Unless someone really really wants this 😄

[natemcmaster]

+1 close. Sounds messy and error-prone.

[Eilon]

Fine with me, wontfix.