New conditionalization scheme

[David L. Andre <DLA at SCRC-TENEX>]

Patches 210.104 and 86.52 implement the new software conditionalization
scheme for lisp machines.  This new scheme will affect users who use the
#+x or #-x reader macros, or the (STATUS FEATURE x) special form (in
other words, any statement herein which refers to #+x refers to all
three constructs).  The changes are:

1.  Read-time conditionalization by site name is no longer possible.
Most notably the forms #+SYM and #+XEROX will no longer work.

2.  Conditionalization by site name can be done at run time, using the
new global variable SITE-NAME.  Typical site names are :MIT, :SCRC, and
:LMI.  The names :EE and :SYM are obsolete.

3.  Read time conditionalization can now be done on software type, with
the #+MIT, #+SYMBOLICS, and #+LMI forms.  These forms will work at any
site.

4.  Read time conditionalization can now be done on hardware type, with
the #+CADR, #+3600, and (I think) #+LAMBDA.  #+LISPM is true on all of
these systems.

5.  It is now possible to do read time conditionalization on supported
network protocols.  Currently there are only two protocols which run on
the lisp machine: #+ETHER, which currently only runs at XEROX PARC, and
#+CHAOS, which runs everywhere else.  Currently, they are mutually
exclusive, but your software shouldn't depend on this fact.