[CHANGE] WeBWorK exercise representations

[Rob Beezer]

If you have any WeBWorK problerms in your project, the following will be of
interest.

When your problems get sent to a WW server for initial processing, you get back
a single file with various different representations of all of your problems.
Now, consistent with how we support other homework systems (MyOpenMath, STACK),
you will get back a single file *for each exercise*. If you use the CLI, this
may all happen under the hood, and Oscar will take care of the details for you.
If you like using the pretext/pretext script you should already be onto how
you will need to adjust, since yuo *like* managing all your associated files
yourself.

We have taken this opportunity to make one other major change that needs to be
reflected in your source.

* Previously, the code tracked problems on its journey from source to server,
back, and into output, via an identifier. Best practice was to set a suggestive
name in the value of #webwork/@xml:id. Otherwise, you got a brittle identifier
like "webwork-832".

* Now, best practice is to put a suggestive name in the value of
#exercise/@xml:id. This is consistent with how we identify Runestone exercises
so exercises can be tracked in a database. Otherwise, you might get a brittle
identifier like "my-great-book-5-2-4-1-1".

If you can turn AI loose on your source, making new @label throughout might be a
good exercise.

[Rob Beezer]

Misspoke on the change, the #exercise (containing a #webwork) should be given a
@label attribute (not an @xml:id attribute). Sorry for the confusion.

Rob