Add support for Fetch and Import-A extensions, and much more

[TheBizzle]

Lord Jeremy, save me, I'm ready for this project to be over.

Here's my summary of what's in this PR:

• Enable default arguments in NLW extensions
• Add a macro for automatically generating the previously-hand-duplicated extension definitions file
• Fix some bugs
• Put polyfills into a dedicated .js file, instead of dumping them as janky Scala strings in Compiler.scala
• Add support for import-pcolors/import-pcolors-rgb
• Expose some stuff in the API (namely, the pure versions of import-* prims)
• Use the latest version JVM NetLogo
• Add infrastructure for docking tests to automatically install JVM NetLogo extensions
• Add infrastructure for mandating that docking tests run with certain extensions that were not necessarily originally included in the model's "Code" tab
• Have export-world and import-world deal with drawing data as base64
• Add the Fetch and Import-A extensions
• Test them
• Update the dumps
• Make the old import-world (and its ilk) throw runtime errors that suggest using Fetch and Import-A instead
• Replace the uses of import-world in the testing infrastructur, and use the extensions now instead

[TheBizzle]

The "branch" tests failed, because (I think) the "pr-merge" tests and the "branch" ones interfered with one another during the import/export portion of TestModels. I reran the "pr-merge" tests afterwards, and they passed, so we can consider all of the tests passed.

[LaCuneta]

This looks awesome! I'm a big fan of pulling out the polyfill code into its own home, and of course of letting people get data in and out of their models in a sensible way. I looked through the code and nothing caught my attention.

I pulled the branch down and tried to run it locally and hit a problem. For the Import-A command tests: Fetch-Integration, ImportPcolors_2D, ImportPcolorsTopologyTest_2D, and RoundTripWithUTF8Chars threw a null reference exception in slurpByType() because the mimeType was null from the call to Files.probeContentType(). I did some brief checking and it looks like Files.probeContentType() might not be reliable across platforms. The other solutions listed there also don't seem great. I tried letting mimeType == null through to be slurped as text, and that failed as well, probably because it wasn't really text. I don't have a great solution, since the alternatives seem like a lot of effort for such a small thing that would only be used in development and testing.

Other than that, I think this is good to go.

[TheBizzle]

:sob:

[TheBizzle]

Alright, I've just pushed a fix to the branch. Everything should (hopefully) be in order now. For maximum compatibility, uninstall your old version of the desktop edition of the Fetch extension. (Deleting your ~/.netlogo folder is a quick and easy way to do this.)

[LaCuneta]

I confirmed the change has things working on my machine now. Waiting for the tests to pass and then I'll merge.