Replace (nearly) all ref cells in the compiler with mutable values (dotnet#8063)

* # This is a combination of 9 commits.
# This is the 1st commit message:

ref -> mutable in more places in the compiler

# The commit message #2 will be skipped:

# Update dependencies from https://github.com/dotnet/arcade build 20191229.1
#
# - Microsoft.DotNet.Arcade.Sdk - 5.0.0-beta.19629.1

# The commit message #3 will be skipped:

# Update dependencies from https://github.com/dotnet/arcade build 20191230.1
#
# - Microsoft.DotNet.Arcade.Sdk - 5.0.0-beta.19630.1

# The commit message #4 will be skipped:

# Update dependencies from https://github.com/dotnet/arcade build 20191231.1
#
# - Microsoft.DotNet.Arcade.Sdk - 5.0.0-beta.19631.1

# The commit message #5 will be skipped:

# Update dependencies from https://github.com/dotnet/arcade build 20200101.1
#
# - Microsoft.DotNet.Arcade.Sdk - 5.0.0-beta.20051.1

# The commit message #6 will be skipped:

# Update dependencies from https://github.com/dotnet/arcade build 20191216.5 (dotnet#8079)
#
# - Microsoft.DotNet.Arcade.Sdk - 1.0.0-beta.19616.5

# The commit message #7 will be skipped:

# dispose fsi at the end of a scripting session (dotnet#8084)
#

# The commit message #8 will be skipped:

# Added static link tests and extended CompilerAssert (dotnet#8101)
#
# * Changed CompilerAssert to static class. Added Compile/Execute methods that take a Compilation description. Added static link tests
# 
# * Hiding compilation description internals
# 
# * Added another test to check for sanity
# 
# * Making a few optional parameters
# 
# * Hiding internals of CompilationReference

# The commit message #9 will be skipped:

# Parameterize product version (dotnet#8031)
#
# * Parameterize Product details
# 
# * fcs
# 
# * Repack pkgdef

* no ilread

Author: cartermp

LanguageService/Tokenizer.fs

@@ -458,7 +458,7 @@ module internal Tokenizer =
         let lineTokenizer = sourceTokenizer.CreateLineTokenizer(lineContents)
         let tokens = ResizeArray<SavedTokenInfo>()
         let mutable tokenInfoOption = None
-        let previousLexState = ref lexState
+        let mutable previousLexState = lexState
 
         let processToken() =
             let classificationType = compilerTokenToRoslynToken(tokenInfoOption.Value.ColorClass)
@@ -471,9 +471,9 @@ module internal Tokenizer =
             tokens.Add savedToken
 
         let scanAndColorNextToken() =
-            let info, nextLexState = lineTokenizer.ScanToken(!previousLexState)
+            let info, nextLexState = lineTokenizer.ScanToken(previousLexState)
             tokenInfoOption <- info
-            previousLexState := nextLexState
+            previousLexState <- nextLexState
 
             // Apply some hacks to clean up the token stream (we apply more later)
             match info with
@@ -519,7 +519,7 @@ module internal Tokenizer =
             classifiedSpans.Add(new ClassifiedSpan(classificationType, textSpan))
             startPosition <- endPosition
 
-        SourceLineData(textLine.Start, lexState, previousLexState.Value, lineContents.GetHashCode(), classifiedSpans.ToArray(), tokens.ToArray())
+        SourceLineData(textLine.Start, lexState, previousLexState, lineContents.GetHashCode(), classifiedSpans.ToArray(), tokens.ToArray())
 
 
     // We keep incremental data per-document.  When text changes we correlate text line-by-line (by hash codes of lines)