Fix PythonAnalyzerSession leak

[AlexanderSher]

Leak has been happening because of the intermediate session has been marked completed, which could trigger a new session while the old one isn't completed. To avoid this, sessions don't reference one another in a form of linked list. Instead, PythonAnalyzer holds reference to the currently-running session and updates the reference for the next session if new request comes before current session is completed.

[AlexanderSher]

Fixes ODE in:

Unhandled Exception: System.ObjectDisposedException: The CancellationTokenSource has been disposed.
 at System.Threading.CancellationTokenSource.ThrowObjectDisposedException()
 at System.Threading.CancellationTokenSource.get_Token()
 at Microsoft.Python.Analysis.Analyzer.PythonAnalyzerSession.<>c__DisplayClass29_0.<StartAnalysis>b__0() in F:\PLS\src\Analysis\Ast\Impl\Analyzer\PythonAnalyzerSession.cs:line 280
 at System.Threading.ExecutionContext.RunInternal(ExecutionContext executionContext, ContextCallback callback, Object state)
--- End of stack trace from previous location where exception was thrown ---
 at System.Threading.Tasks.Task.ExecuteWithThreadLocal(Task& currentTaskSlot)
--- End of stack trace from previous location where exception was thrown ---
 at Microsoft.Python.Core.TaskExtensions.<>c.<DoNotWaitThreadContinuation>b__4_0(Object s) in F:\PLS\src\Core\Impl\Extensions\TaskExtensions.cs:line 79
 at System.Threading.ExecutionContext.RunInternal(ExecutionContext executionContext, ContextCallback callback, Object state)
--- End of stack trace from previous location where exception was thrown ---
 at System.Threading.ThreadPoolWorkQueue.Dispatch()

[jakebailey]

master peaks at some 4GB with tensorflow, and sometimes drops to ~1.3GB at the end, now it peaks at completion at ~1.3GB at the end. Night and day.