Make Str.chomp between 10x and 100x faster

- 10x if nothing needed to be done
- 100x if a newline needed to be removed

Based again on the observation that nqp::substr will return self for
whole strings, so the whole logic of .chomp could be caught in a single
nqp::substr.  Only worry is that:

    nqp::iscclass(nqp::const::CCLASS_NEWLINE,self,-1)

may actually check out of bounds.  If so, suggest that should be fixed
in the VM, as making .chomp 100x faster for newline removal looks pretty
important to me when using lines()

src/core.c/Str.pm6

@@ -105,11 +105,12 @@ my class Str does Stringy { # declared in BOOTSTRAP
     }
 
     method chomp(Str:D: --> Str:D) {
-        nqp::if(
-          (nqp::isge_i((my int $chars = nqp::sub_i(nqp::chars($!value),1)),0) #?js: NFG
-            && nqp::iscclass(nqp::const::CCLASS_NEWLINE,$!value,$chars)), #?js: NFG
-          nqp::p6box_s(nqp::substr($!value,0,$chars)), #?js: NFG
-          self
+        nqp::substr(
+          self,
+          0,
+          nqp::chars(self) - nqp::iscclass(                       #?js: NFG
+            nqp::const::CCLASS_NEWLINE,self,nqp::chars(self) - 1  #?js: NFG
+          )
         )
     }