Fix sub createRandomPoints so that when test_at points are included, …

[taniwallach]

Fix sub createRandomPoints so that when test_at points are included it still generates num_points random points in addition to the provided points, as intended. This fixes the issue reported in #472 so that the test_at works as documented in https://webwork.maa.org/wiki/Context_flags .

[dpvc]

Looks good.

[taniwallach]

The easiest way to see how this changes the behavior is to set diagnostics=>1 inside a cmp() call where the relevant MathObject Formula has test_at set. Then during grading a table about the test points and what occurred at each is shown.

A sample PG file which sets 8 test_at points and has that flag set is below.

Before the patch, when graded - that problem would only test at the 8 given points.
After the patch - those 8 points and 5 random points are tested.

---

DOCUMENT();

loadMacros(
  "PGstandard.pl",
  "MathObjects.pl",
);

Context("Numeric");
Context()->variables->are(r =>'Real', t =>'Real',p =>'Real');

$int  = Compute("(r^2)*sin(p)*cos(t)");

$int->{test_at} = [ [ 1, 1, 1],
                 [ 1, 1,-1],
                 [ 1,-1, 1],
                 [ 1,-1,-1],
                 [-1, 1, 1],
                 [-1, 1,-1],
                 [-1,-1, 1],
                 [-1,-1,-1],
               ];

#################################################################################

Context()->texStrings;
BEGIN_TEXT
Enter \( $int \) below:

$BCENTER
\{ $int->ans_rule(20)\}
$ECENTER

END_TEXT
Context()->normalStrings;

ANS( $int->cmp( diagnostics=>1 ) );

ENDDOCUMENT();

[taniwallach]

@drgrice1 Do you mind doing a quick test and merging?

[drgrice1]

I can verify that this works as stated. Without the pull request the diagnostics shows only 8 test points were used. With the pull request 5 other random test points are also used.

Looks good.