Incorrect domain of `t` used in `stats/base/dists/normal/mgf` across benchmark, test and examples

[manit2004]

For the mgf function in normal distribution mgf( t, mu, sigma) t is allowed to take any value within (-inf, +inf) but in the test, benchmarks and examples the domain used is [0, +inf) except the C benchmark file.

• For C benchmark in stats/base/dists/normal/mgf/benchmark/c/benchmark.c at line 105 the domain of t is correct.

	for ( i = 0; i < 100; i++ ) {
		t[ i ] = random_uniform( -100.0, 100.0 );

• For JS benchmarks in stats/base/dists/normal/mgf/benchmark/benchmark.js at line 76, line 45 and stats/base/dists/normal/mgf/benchmark/benchmark.native.js at line 54 the domain of t is [0,1)

	t = uniform( 100, 0.0, 1.0, {
		'dtype': 'float64'
	});

• For both C and JS examples in stats/base/dists/normal/mgf/examples/index.js at line 30 the domain is [0,10) and in stats/base/dists/normal/mgf/examples/c/example.c at line 37 the domain is [0,1)

JS file

var t = uniform( 10, 0.0, 10.0, opts );

C file

t = random_uniform( 0.0, 1.0 );

• For the Julia test fixture runner in stats/base/dists/normal/mgf/test/fixtures/julia/runner.jl at line 74, line 80 and line 86 the value x (which corresponds to t) is generated only from [0, +inf), missing coverage of negative values which are valid inputs.

x = rand( 1000 ) .* 20.0;
...
x = rand( 1000 ) .* 20.0;
...
x = rand( 1000 ) .* 10.0;

• For the README.md examples in stats/base/dists/normal/mgf/README.md at line 142 the JS example uses [0,10) and at line 233 the C example uses [0,1) for the domain of t.

JS example

var t = uniform( 10, 0.0, 10.0, opts );

C example

t = random_uniform( 0.0, 1.0 );