Base implementation of Dynamix, Tim

[molenzwiebel]

This PR contains the following changes:

• A new metalang, tim, which is the result of running a Dynamix specification on an input AST. This is a CPS-based intermediate representation language. The tim metalang includes a simple Stratego interpreter, but the intention is that this is replaced with a more performant optimizing compiler in a follow-up master thesis.
• The dynamix metalang has been split into dynamix (grammar, static analysis, "compilation"/spec merging) and dynamix_runtime (stratego interpreter for running compiled/merged specifications).
• Enable multi-file analysis for dynamix, and support running the statix spec inside the concurrent interpreter.
• Tasks/menus for compiling the dynamix specification in a project to an ATerm representation, and using the dynamix_runtime for executing the spec on an input AST.
• All languages (tim/dynamix/dynamix_runtime) have been namespaced into folders, their sort/constructor names prefixed (T/M respectively) and stratego rules prefixed (tim-, dx- respectively).
• The tim/dynamix/dynamix_runtime spoofax2 projects use source imports to depend on each other.
• Actual signature generation for dynamix is implemented, instead of simple stubs.

Notes/issues that would be nice for you to take a look at:

• The ReadDynamixSpecTaskDef is possibly a bit over the top and could be changed to just passing a ResourcePath, and having the reading be done in the dynamix runtime instead, but I wasn't entirely sure if that would work properly with PIE.
• Menu items are included for running the Dynamix specification. However, the continuous versions of these seem to not work. Possibly due to the fact that the output is written to a file.
• Embedding dynamix_runtime should be complete (largely copied from the rv32im code), but I'm not entirely certain if is correct. I am not sure whether a standalone eclipse build will work.
• Passing options to the DynamixRuntime is currently done through createDynamixRuntimeConfig in ExecuteDynamixSpecificationTaskDef.java.mustache. This works, but there might be a more idiomatic approach?
• The dynamix_runtime "language" is currently in the metalang folder, but it might be more appropriate to put it elsewhere (next to rv32im?).
• The tim stratego interpreter is currently bundled with the metalang (and not split into tim_runtime or something similar). Currently done because it works and as mentioned this shouldn't be the long-term solution for running tim files.
• Source imports for ESV files don't seem to work. Try uncommenting the tim/colors import in the dynamix Main.esv.
• (Spoofax 2) versions of all the languages are set to 0.1.0-SNAPSHOT. Is that fine?
• All stratego files have commented out type signatures for str2. Ideally the files are ported over the str2, but I had some issues with getting Stratego 2 to work properly in Spoofax 2 so we're back on the old stratego for now.

Other notes/TODOs:

• Spoofax 3 doesn't yet support configuring the Statix solver mode. @AZWN, I see that #108 is open for that?
• Current statix spec becomes very slow very fast, big culprit here is the checking for overlaps in rule patterns. This is partially on my end, and partially something which might be due to an issue in the Statix solver?
• The Statix spec seems to work for the concurrent solver (tested in Spoofax 2), but the concurrent solver is significantly slower. The incremental solver seems to fail. I sent a bundled version to @AZWN, hopefully he has some time to take a look at some point.

[AZWN]

To quickly react to your last points involving the concurrent solver: When your specification semantically works with the concurrent solver, and there is no type-directed transformation, I think it is fine to use the traditional solver for now, and switch to the concurrent solver later. I'll add a TODO to that PR.

[AZWN]

Exactly 11 months ago, I posted a comment about enabling the concurrent solver. I don't remember the discussion fully anymore, but the referred PR has landed, so technically, enabling the concurrent solver might be possible.