Use encoding and CR-aware concatenation

The old `cat` method used here does not consider encodings or
code ranges, instead just slamming the given bytes directly into
the existing buffer. This causes problem when those bytes or the
encoding they came from does not match the existing encoding or
code range, leading to issues like #247.

The fix here uses a code range-aware method for doing the bytes
concatenation, and resolved the related bug.

Fixes #247

Author: headius

src/main/java/org/jruby/rack/ext/Input.java

@@ -29,6 +29,7 @@
 import org.jruby.rack.RackEnvironment;
 import org.jruby.rack.servlet.RewindableInputStream;
 import org.jruby.rack.util.ExceptionUtils;
+import org.jruby.util.StringSupport;
 
 /**
  * Native (Java) implementation of a Rack input.
@@ -143,7 +144,7 @@ public IRubyObject read(final ThreadContext context, final IRubyObject[] args) {
             if ( bytes != null ) {
                 if ( buffer != null ) {
                     buffer.clear();
-                    buffer.cat(bytes);
+                    buffer.catWithCodeRange(new ByteList(bytes, false), StringSupport.CR_UNKNOWN);
                     return buffer;
                 }
                 return context.runtime.newString(new ByteList(bytes, false));