Fix just introduced possible old-http bug

My last patch changed httprangereader.read to read only the specified
amount of data from the connection, to prevent it from returning more
than what was asked.

I just realized that this could lead to the connection not being closed.
In practice, it looks like the connection is closed just fine, but it's
probably safer to read everything and then return only what's necessary.

mercurial/httprangereader.py

@@ -1,25 +1,28 b''
 # httprangereader.py - just what it says
 #
 # Copyright 2005 Matt Mackall <[e-mail]>
 #
 # This software may be used and distributed according to the terms
 # of the GNU General Public License, incorporated herein by reference.
 
 import byterange, urllib2
 
 class httprangereader(object):
     def __init__(self, url):
         self.url = url
         self.pos = 0
     def seek(self, pos):
         self.pos = pos
     def read(self, bytes=None):
         opener = urllib2.build_opener(byterange.HTTPRangeHandler())
         urllib2.install_opener(opener)
         req = urllib2.Request(self.url)
         end = ''
         if bytes:
             end = self.pos + bytes - 1
         req.add_header('Range', 'bytes=%d-%s' % (self.pos, end))
         f = urllib2.urlopen(req)
-        return f.read(bytes)
+        data = f.read()
+        if bytes:
+            data = data[:bytes]
+        return data