Investigate usage of bytecode interpreter

[JaroslavTulach]

Truffle is moving towards byte code interpreters. It would help Enso as well for example to:

• eliminate deep stacktraces in the interpreter
• have a standard format of .ir caches

Goal: Investigate what it would take to use bytecode interpreter in Enso.

Opportunity

Generating the bytecode is going to be something completely different than working with AST nodes. It is a great opportunity to remove the need of IR and (also) move towards incremental compilation. Let's generate the bytecode directly from Tree and avoid using IR all together. Of course, to make that happen we need to align the Tree and IR to have similar structure. @kazcw's work like:

• Move TypeSignature into field of Function #11364
• Replace private modifier node with field on supporting types #11346
• Distinguish assignment/thunk by statement context #11324
• & co.

is thus quite important in the bytecode interpreter effort.

[JaroslavTulach]

oracle/graal#9556 is still opened. Once it gets integrated we should again some experience with the bytecode interpreters. To do so, I'd like to try implementing a bytecode interpreter for ECMA-55_1st_edition_january_1978.pdf - the Minimal BASIC specification from 1978.