cborutil: fix streamencode() to handle subtypes

Otherwise the template filter 'cbor' could crash because of bytes subclass:

  ValueError: do not know how to encode
  <class 'mercurial.encoding.safelocalstr'>

mercurial/utils/cborutil.py

@@ -214,6 +214,14 b' def streamencode(v):'
     fn = STREAM_ENCODERS.get(v.__class__)
 
     if not fn:
+        # handle subtypes such as encoding.localstr and util.sortdict
+        for ty in STREAM_ENCODERS:
+            if not isinstance(v, ty):
+                continue
+            fn = STREAM_ENCODERS[ty]
+            break
+
+    if not fn:
         raise ValueError('do not know how to encode %s' % type(v))
 
     return fn(v)

tests/test-template-functions.t

@@ -1562,6 +1562,20 b' json filter takes input as utf-8b:'
   $ HGENCODING=ascii hg log -T "{'`cat latin1`'|json}\n" -l1
   "\udce9"
 
+cbor filter is bytes transparent, which should handle bytes subtypes
+as bytes:
+
+  $ HGENCODING=ascii hg log -T "{branch|cbor}" -r0 \
+  > | "$PYTHON" "$TESTTMP/decodecbor.py"
+  [
+   '?'
+  ]
+  $ HGENCODING=latin-1 hg log -T "{branch|cbor}" -r0 \
+  > | "$PYTHON" "$TESTTMP/decodecbor.py"
+  [
+   '\xe9'
+  ]
+
 utf8 filter:
 
   $ HGENCODING=ascii hg log -T "round-trip: {branch|utf8|hex}\n" -r0