DYN-4231 Fast Path for sweep during FullGC

[saintentropy]

Purpose

https://jira.autodesk.com/browse/DYN-4231

The purpose of this PR is to optimize the Sweep pass for the FullGC pass that occurs after any change to the graph that causes re-execution. Specifically this optimization does the rolling:

• introduces a specific Call method to the Executive (CallDispose) tailored specifically for dispatching Dispose calls vs allowing these calls from the GC to pass through the generic Callr. Additionally it introduces a specific Dispatch method to the Callsite ('DispatchDispose) also tailored to Dispatch Dispose calls vs Dispatch. In both case the normal overhead of the CallrandDispatchis much higher and .NET memory/GC intensive than the actualDispose` calls.
• Adds caching based on object type where possible as the GC is often collecting similar types repeatedly

In testing a graph that had ~150000 items to collect in the sweep pass the time was reduced from 2s to .5s and the .NET memory allocation from 800mb to 16mb.

Todo is to validate with testing include DS object dispose.

Declarations

Check these if you believe they are true

• The codebase is in a better state after this PR
• Is documented according to the standards
• The level of testing this PR includes is appropriate
• User facing strings, if any, are extracted into *.resx files
• All tests pass using the self-service CI.
• Snapshot of UI changes, if any.
• Changes to the API follow Semantic Versioning and are documented in the API Changes document.
• This PR modifies some build requirements and the readme is updated

Reviewers

TBD

FYIs

@jasonstratton @QilongTang

[aparajit-pratap]

Why have these lines been removed?

[saintentropy]

This was not necessary with the new route and would be a memory leak. Also FYI in Dynamo today it does seem to have a memory leak regardless. We always add 3 items but only ever Pop two.

[sm6srw]

Many thanks @aparajit-pratap for your insights. I have been trying to break this in various ways, unsuccessfully so far. It works well. In many cases where it currently takes 10-15 times longer to dispose objects than it takes to create them it now takes about the same time to create and dispose. I am also seeing about 20% less memory being used.

[sm6srw]

@saintentropy You have 4 test failures now.

[aparajit-pratap]

Why is the context (first arg) passed as null?

[saintentropy]

Only because for this execution path it was not referenced.