Move logic for verifying that the caret can map into the subject buffer into StartNewModelComputation

After calling nextHandler(), backspace completion should verify that the caret can still be mapped into our subject buffer before starting a session.

Fixes #3789.

Add CscCore to CrossPlatform.sln

Make NuGet sources more deterministic

Async loading with property pages

Updates the mono toolset to mono built from source at tag
mono-4.2.0.207.

Updates the mono toolset to mono built from source at tag 4.2.0-207.

This is needed to unblock the core change to fix issue #4183.

project.json Fixes

Added a clear entry to our package sources to ensure builds on developer machines weren't inadvertently affected by local settings.

Update the project.lock.json for FakeSign and upload a new nuget zip

#load directives...

It isn't needed, as the Workspace will notice changes in the Compilation
due to #loads.

Tidy up nuget artifacts for ResultProviders...

Fixes a subtle bug in branch optimizer.

Making stack scheduling optimizations more robust and fixing #4828

Performance improvements for the v2 analyzer driver

Using correct case in-sensitive comparer
Fixes #4782

Address PR feedback

Remove check for 'IgnoreCase In CSharpAddImportCodeFixProvider we
explicitly override IgnoreCase to be false.  Therefore, we will never use
a case in-sesitive comparer.

Prioritize uncommenting single line comments over multi-line comments.

Fixes: #548

1) Reduce number allocations in the analyzer driver that show up perf traces for Picasso.
2) Currently we maintain a cache of CompilationWithAnalyzers for projects being actively analyzed. This is causing the VM to rise and also forcing GC to kick in at different phases causing regressions at various scenarios. The fix is to just create a new CompilationCompilationWithAnalyzers instance for each document/project analysis request.

Add support for multi-dimensional arrays of rank 1 to C# compiler.

Fixes #4924.

Start outlining span after all the whitespace on the target token.

list exceptions from the doc comments when showing quick info

"branch over branch" optimization eliminates chains of trivial basic blocks from the graph, which is correct as long as trivial blocks are not reachable individually from outside via branches.
The "same as next" optimization can make a block trivial and if it was targeted by a branch, break the assumption, leading to a rare situation of some branches having stale offsets.

The fix makes sure if a block becomes trivial, its labels are moved to the next nontrivial block.

Fixes #4838
Fixes #4839