[Rd] attributes on symbols

[Radford Neal]

Attributes on symbols seem like a bad idea to me.  An additional
obscure part of the global state seems undesirable.  I can't see how
any use of them would be preferrable to storing an environment in some
global variable, in which the same information could be recorded.

Note that the attributes on symbols are not saved when the workspace
is saved with q("yes").

   Radford Neal