Add interp tool like TestRig

[parrt]

An interpreted version of TestRig, which only ran Java-based (and compiled) parser/lexers. This uses java-based interp to process any grammar (ignoring actions of course).

Needed by new tools: https://github.com/antlr/antlr4-tools

$ java org.antlr.v4.gui.Interpreter \
    /Users/parrt/sample/TParser.g4 \
    /Users/parrt/sample/TLexer.g4 \
    expr -profile /tmp/T-profile.csv -tokens /tmp/foo.txt

$ java org.antlr.v4.gui.Interpreter /Users/parrt/sample/JSON.g4 json -profile /tmp/json.csv /tmp/foo.json
Rule,Invocations,Time,Total k,Max k,Ambiguities,DFA cache miss
obj:0,2,0.030417,2,1,0,2
obj:1,1,1.317417,2,2,0,2
arr:2,3,0.034917,3,1,0,2
arr:3,1,0.101917,2,2,0,2
value:4,6,0.341458,6,1,0,3

@kaby76 @KvanTTT @hzeller @mike-lischke @ericvergnaud might be interested. Then I'll hook into python script which installs java/antlr.

$ java org.antlr.v4.gui.Interpreter /Users/parrt/sample/JavaParser.g4 /Users/parrt/sample/JavaLexer.g4 compilationUnit -profile /tmp/a.csv -trace
class T { int i; }
enter   compilationUnit, LT(1)=class
enter   typeDeclaration, LT(1)=class
enter   classDeclaration, LT(1)=class
consume [@0,0:4='class',<9>,1:0] rule classDeclaration
enter   identifier, LT(1)=T
consume [@2,6:6='T',<128>,1:6] rule identifier
exit    identifier, LT(1)={
enter   classBody, LT(1)={
consume [@4,8:8='{',<80>,1:8] rule classBody
enter   classBodyDeclaration, LT(1)=int
...
exit    compilationUnit, LT(1)=<EOF>
$ head /tmp/a.csv
Rule,Invocations,Time (ms),Total k,Max k,Ambiguities,DFA cache miss
compilationUnit:0,1,0.30525,1,1,0,1
compilationUnit:1,1,0.578416,1,1,0,1
compilationUnit:2,2,0.634208,2,1,0,2
compilationUnit:3,1,1.662042,1,1,0,1
packageDeclaration:4,0,0.0,0,0,0,0
importDeclaration:5,0,0.0,0,0,0,0
importDeclaration:6,0,0.0,0,0,0,0
typeDeclaration:7,1,0.0925,1,1,0,1
typeDeclaration:8,1,0.086917,1,1,0,1

[parrt]

Needed by https://github.com/antlr/antlr4-tools. Can discuss antlr/antlr4-tools#2