JDK 9 experiments around module-info, automatic modules, etc

Try something like CloudBalancingHelloWorld with just a dependency on optaplanner-core in 3 scenario's:
1) the HelloWorld compiled and running on JDK 8
2) the HelloWorld without a module-info.java file compiled and running on JDK 9
3) the HelloWorld with a module-info.java file compiled and running on JDK 9

combine this with a local fork of optaplanner-core (and potentially drools-compiler, drools-core, kie-api and kie-internal-api too) in the following forms:
A) (current form) optaplanner-core build on JDK 8 with no manifest entry
B) optaplanner-core build on JDK 8 with a (automatic) module-name manifest entry "org.optaplanner.core".
C) optaplanner-core build on JDK 9 with a (automatic) module-name manifest entry "org.optaplanner.core".
D) optaplanner-core build on JDK 9 with -release 8 (so it works on JDK 8 too) with module-info.java

Write a 1-page google doc for MarkP, Edson, Kris and Geoffrey (but do share with bsig-all too) with:

• a short paragraph clear explanation of module-info vs manifest entry vs nothing
• an objective list of how each combination worked (or didn't)
• add advice on what you liked/disliked, but clearly mark it as an opinion
  • changing the background of pro's and con's to light green and light red helps for people skimming through the doc
• propose a plan to become JDK9-friendly, for engines and workbenches. Our primary focus in the engines atm.

Tip: speed up dev time by go into the hello world benchmark config and turn down the 120 seconds to just 10 or so. Maybe also reduce or disable the warmup (see docs) which defaults to 30 seconds.