STEP-6.2 CHANGES: new/changed keys to make sudirectories more useful

b Base-directory [startup-dir]: 

    b <ret>                 reset base to startup directory
    b path <ret>            set base directory to path

This changes the base of the path used for the "e" (experiment
subdirectory) key only.  "base-directory" is stored as an absolute path.


e Experiment-subdirectory [last-thing-loaded-dir]:

    e <ret>                 reset working dir to last dir loaded from
    e path <ret>            set working dir to base/path

This key is used to change the working directory, and to then print
out a description of experiments in the new working directory.  To do
this, we first cd to the base directory, and then we cd to the
indicated experiment subdirectory.  We then print out any README file
contained in this directory, followed by the first line of each file
in that directory that has the extension ".exp", ordered from oldest
to newest.  The newest ".exp" file in this directory becomes the new
default for the "l" key.

Note: the "/" key is no longer used to change directories.


l Load-file [last-exp-loaded or newest-exp]:

    l <ret>                 load default from current directory
    l filename <ret>        load "filename" from current directory

This key has not changed, except for the default, which is set to
"newest-exp" whenever we change to a new working directory (see
"e" key).

NOTE: If a non-absolute-path file to be loaded is not found in the
current directory, we look for it in turn in each of the directories
listed in the environment variable STEP_INPUTS.  This variable is set
up by the standard "im.cshrc" file.  This applies to the "load"
command as well.  For example, "load 3x3.hood" in an experiment will
now load this file from a standard place (currently "/cam8/common").
As another example, the "diags.exp" experiment can now be loaded from
anywhere.  Before loading any file via STEP_INPUTS, we make the
directory containing that file the new working directory -- it is
changed back when we're done loading that file.  This can happen
recursively. 


d Directory-listing [base-dir]:

    d <ret>                 list base directory
    d .<ret>                list working directory
    d dirname <ret>         list subdirectory of working directory
    d path/*.pat.*  <ret>   list path/*.pat.* starting from working dir

    This just uses "ls" with the given argument (or the default argument).


Normal usage:

o Start step in a directory containing one subdirectory per
  experiment.  This sets the base directory.  Subdirectories might
  have names like "anneal", "life", etc.  

o To list available experiments, type "d<ret>".  This lists the base
  directory. 

o To work with experiment(s) in the "anneal" subdirectory, type
  "e anneal<ret>".  This will cd to the anneal subdirectory and print
  some lines of documentation.

o To load the newest experiment in the "anneal" directory, type "l<ret>".
  To load some other experiment from this directory, type "l name<ret>". 

o To switch to the "life" experiments, type "e life<ret>".  To run
  the newest life experiment, type "l<ret>".

etc.
