Ensure that an EnvDTE.CodeParameter.Name returns the correct value for parameters defined in metadata

Fix assert failure for FileThatCannotBeDecoded under default codepage 936

Ensure that an EnvDTE.CodeParameter.Name returns the correct value for parameters defined in metadata

The issue is that the symbol display system will not return a name for a parameter unless
SymbolDisplayParameterOptions.IncludeName is set. This behavior seems like it might ultimately be
a bug in symbol display as that option is intended to control the display the parameters owned
by the targeted symbol, but probably shouldn't affect the display of the targeted symbol. However,
rather than updating symbol display (which would potentially have larger downstream impact), we'll
just include the option in this case.

Theme the rename dashboard's "New name:" TextBlock

Cache test results locally

Remove LTTNG warnings on Linux

change all [Fact] private tests to be public

Fix TypeChar handler in Interactive Window

Code Model: GetTypeSymbolFromFullName should not return unpredictable results when the type name was in an unsupported format

Do not run Add-Using on syntax errors.

Enforce that WpfFact tests must run serially

Fixes #6721

Move handling of TypeChar in InteractiveWindow from PreLanguageCommandFilter to PreEditorCommandFilter

Fix test issues

For now, WpfTestCase, when running tests, sets various bits of global
static state. Because of this, it's not safe to let two tests interleave
on the same test thread, and so we should explicitly prevent this.

This change we should revert in the hypothetical future where we can
run multiple WpfFact threads in parallel.

This fixes as couple of test issues:

- CultureAttributeMismatch: this was a flaky suite whose flakiness was
fixed via the PEVerify changes.  The hardened rules revealed the suite
was setup to consistently fail verification because it didn't fully load
it's dependencies into the Compilation object.  This behavior is
actually expected, we want to get the compiler errors that result from
such a compilation.  Hence we disable verification in those cases.
- Return07_2: This was directly fixed by the PEVerify hardenning.

closes #4300
closes #5866

The LTTNG library version used in CoreCLR requires LTTNG_HOME be set
else it issues a warning on process startup.  This was cluttering up our
linux build output.  Fixing this by exporting the value.

Fix `InteractivePasteCommandHandler` to handle line-copy and box-copy

Fix handling of abstract/virtual in structs

Throw an exception in Code Model when an attempt is made to get a metadata name for a type parameter

This should never happen, but there is a case where it was (https://github.com/dotnet/roslyn/issues/7179).
That particular scenario has been corrected but we'll throw here in case it every happens again. Otherwise,
we could stack overflow.

Fix Code Model's GetTypeFromFullName to better handle situations where unsupported name formats are passed

`GetTypeFromFullName` first attempts to treat the specified type name as a metadata name. Failing
that, it parses the type name and tries to speculatively bind it. However, it was entirely possible
that a type name could be provided that we don't support, such as an assembly-qualified name. So, we
now check to see if the parsed type name includes an skipped token trivia (e.g. a trailing "Foo, 0.0.0.0")
and does not attempt to bind in that case. Otherwise, speculative binding could return something
non-sensical.

Don't parse out a local declaration where there clearly isn't one.

Fixes: #4014

Remove BuildTask ref from VS Setup project

If we see:

```c#
  Identifier1
  Identifier2();
```

Don't parse that out as a declaration of a local called "Identifier2".