[CT-2459] Move ParsedNode.patch and Macro.patch to respective PatchParsers

[MichelleArk]

Currently a couple node types (Macro, ParsedNode) implement their own patch method which accepts a patch (either ParsedNodePatch or ParsedMacroPatch) and are responsible for applying the contents of the patch to the node attributes. The PatchParsers then call these node patch methods once after constructing a patch.

This splits the overall patching responsibilities into creating a patch and applying it across the PatchParsers and Nodes (which is not really that clean of a split in reality because NodePatchParser implement its own patch_node_config method that applies patches to node configs specifically). Most of the complexity is in creating a patch rather than 'applying' it, and this split requires changes to patching functionality in multiple places each time a new patch attribute is introduced. Additionally, this makes the PatchParsers difficult to test because applying actual result of their processing is mocked out.

Let's move the Node.patch methods into the respective PatchParsers instead to consolidate overall patching logic. We should also consider creating a separate ParsedModelPatch for model-specific configs that are patched by ModelPatchParser if possible (could be tricky with the inheritance currently in place).