[llvm] [lit] Support wildcard in --xfail-not option (PR #151191)

[Joel E. Denny via llvm-commits]

================
@@ -2175,6 +2175,8 @@ def parseIntegratedTestScript(test, additional_parsers=[], require_script=True):
     assert parsed["DEFINE:"] == script
     assert parsed["REDEFINE:"] == script
     test.xfails += parsed["XFAIL:"] or []
+    if test.xfails and test.skip_xfail:
+        return lit.Test.Result(Test.SKIPPED, "skipping XFAIL tests")
----------------
jdenny-ornl wrote:

For consistency with other usage, it seems like the result code ought to be `Test.EXCLUDED`, which seems to be used for cases when the user has excluded tests up front.  `Test.SKIPPED` seems to be used for cases when a particular run didn't manage to reach a test even though it was not excluded by the user up front.  For example, in `llvm/utils/lit/lit/reports.py`, see `XunitReport._get_skip_reason`, which has:

```
if code == lit.Test.EXCLUDED:
    return "Test not selected (--filter, --max-tests)"
if code == lit.Test.SKIPPED:
    return "User interrupt"
```

Maybe we should also rename the option to `--exclude-xfail` for consistency?  Sorry for not thinking of this before I proposed `--skip-xfail`.  I had not thought about result codes yet.

https://github.com/llvm/llvm-project/pull/151191