Better support for stack machine decompilation

Right now binja's core deompilation capacity is more focused on handling register machines like x86 and arm. It would be great if we can also handle stack machines, e.g., Java/.NET/Python/WASM bytecode.

Right now, technically we can represent the operations of a stack machine using our IL, but the decompiler output is often times not optimal. We probably need to do some specific handling to improve the decompilation of a stack machine.

Note this is related to https://github.com/Vector35/binaryninja-api/issues/742 but not the same

Note, to have good support for Java/.NET, there are various other language specific issues beyond this that need to be addressed. WASM might be a good starting point in this sense.

Also this is unlikely to happen in the near future, however, I am creating this issue so that people can refer to it in discussion

No file, test, or entry point is named. Read the related issue 742 and compare the current IL representation with decompilation of stack-machine bytecode, using WASM as the suggested starting point. Done would mean materially better, more optimal decompiler output for stack machines, with language-specific gaps tracked separately.