Adjust conversion of AbstractFile's name to TypeName

The current optimized version tries to avoid temporary strings.
But it doesn't achieve this for classes based by jrt:// (or any
`NioPath`, as the call to `AbstractFile.fileName` internally
constructs a string each time.

This commit uses `.name` (which is a `lazy val`).

src/compiler/scala/tools/nsc/symtab/SymbolLoaders.scala

@@ -199,15 +199,16 @@ abstract class SymbolLoaders {
     }
   }
   private def nameOf(classRep: ClassRepresentation): TermName = {
-    while(true) {
-      val len = classRep.nameChars(nameCharBuffer)
-      if (len == -1) nameCharBuffer = new Array[Char](nameCharBuffer.length * 2)
-      else return newTermName(nameCharBuffer, 0, len)
+    val name = classRep.name
+    val nameLength = name.length
+    if (nameLength <= nameCharBuffer.length) {
+      name.getChars(0, nameLength, nameCharBuffer, 0)
+      newTermName(nameCharBuffer, 0, nameLength)
+    } else {
+      newTermName(name)
     }
-    throw new IllegalStateException()
   }
-  private var nameCharBuffer = new Array[Char](256)
-
+  private val nameCharBuffer = new Array[Char](512)
 
   /**
    * A lazy type that completes itself by calling parameter doComplete.

src/compiler/scala/tools/nsc/util/ClassPath.scala

@@ -204,17 +204,6 @@ object ClassPath {
 trait ClassRepresentation {
   def fileName: String
   def name: String
-  /** Low level way to extract the entry name without allocation. */
-  final def nameChars(buffer: Array[Char]): Int = {
-    val ix = fileName.lastIndexOf('.')
-    val nameLength = if (ix < 0) fileName.length else ix
-    if (nameLength > buffer.length)
-      -1
-    else {
-      fileName.getChars(0, fileName.lastIndexOf('.'), buffer, 0)
-      nameLength
-    }
-  }
   def binary: Option[AbstractFile]
   def source: Option[AbstractFile]
 }