fix: update hypot to follow the IEEE 754-2019 standard

lib/node_modules/@stdlib/math/base/special/hypot/README.md

@@ -52,7 +52,7 @@ h = hypot( -0.0, -0.0 );
 // returns +0.0
 ```
 
-If either argument is `NaN`, the function returns `NaN`.
+If either argument is `NaN` and the other argument is not `+-Infinity`, the function returns `NaN`.
 
 ```javascript
 var h = hypot( NaN, 12.0 );

lib/node_modules/@stdlib/math/base/special/hypot/docs/repl.txt

@@ -2,7 +2,8 @@
 {{alias}}( x, y )
     Computes the hypotenuse avoiding overflow and underflow.
 
-    If either argument is `NaN`, the function returns `NaN`.
+    If either argument is `NaN` and the other argument is not `+-Infinity`,
+    the function returns `NaN`.
 
     Parameters
     ----------

lib/node_modules/@stdlib/math/base/special/hypot/lib/main.js

@@ -49,12 +49,14 @@ var sqrt = require( '@stdlib/math/base/special/sqrt' );
 */
 function hypot( x, y ) {
 	var tmp;
-	if ( isnan( x ) || isnan( y ) ) {
-		return NaN;
-	}
+
+	// If one of the arguments is `+-infinity`, return `+infinity` even if the other argument is `NaN` (IEEE 754-2019)...
 	if ( isInfinite( x ) || isInfinite( y ) ) {
 		return PINF;
 	}
+	if ( isnan( x ) || isnan( y ) ) {
+		return NaN;
+	}
 	if ( x < 0.0 ) {
 		x = -x;
 	}

lib/node_modules/@stdlib/math/base/special/hypot/src/main.c

@@ -37,12 +37,14 @@ double stdlib_base_hypot( const double x, const double y ) {
 	double tmp;
 	double a;
 	double b;
-	if ( stdlib_base_is_nan( x ) || stdlib_base_is_nan( y ) ) {
-		return 0.0 / 0.0; // NaN
-	}
+
+	// If one of the arguments is `+-infinity`, return `+infinity` even if the other argument is `NaN` (IEEE 754-2019)...
 	if ( stdlib_base_is_infinite( x ) || stdlib_base_is_infinite( y ) ) {
 		return STDLIB_CONSTANT_FLOAT64_PINF;
 	}
+	if ( stdlib_base_is_nan( x ) || stdlib_base_is_nan( y ) ) {
+		return 0.0 / 0.0; // NaN
+	}
 	a = x;
 	b = y;
 	if ( a < 0.0 ) {

lib/node_modules/@stdlib/math/base/special/hypot/test/test.js

@@ -72,10 +72,22 @@ tape( 'the function returns `+infinity` if either argument is `+-infinity`', fun
 	h = hypot( NINF, NINF );
 	t.strictEqual( h, PINF, 'returns expected value' );
 
+	h = hypot( NaN, PINF );
+	t.strictEqual( h, PINF, 'returns expected value' );
+
+	h = hypot( PINF, NaN );
+	t.strictEqual( h, PINF, 'returns expected value' );
+
+	h = hypot( NINF, NaN );
+	t.strictEqual( h, PINF, 'returns expected value' );
+
+	h = hypot( NaN, NINF );
+	t.strictEqual( h, PINF, 'returns expected value' );
+
 	t.end();
 });
 
-tape( 'the function returns `NaN` if either argument is `NaN`', function test( t ) {
+tape( 'the function returns `NaN` if either argument is `NaN` but not `+-infinity`', function test( t ) {
 	var h;
 
 	h = hypot( NaN, 3.14 );

lib/node_modules/@stdlib/math/base/special/hypot/test/test.native.js

@@ -81,10 +81,22 @@ tape( 'the function returns `+infinity` if either argument is `+-infinity`', opt
 	h = hypot( NINF, NINF );
 	t.strictEqual( h, PINF, 'returns expected value' );
 
+	h = hypot( NaN, PINF );
+	t.strictEqual( h, PINF, 'returns expected value' );
+
+	h = hypot( PINF, NaN );
+	t.strictEqual( h, PINF, 'returns expected value' );
+
+	h = hypot( NINF, NaN );
+	t.strictEqual( h, PINF, 'returns expected value' );
+
+	h = hypot( NaN, NINF );
+	t.strictEqual( h, PINF, 'returns expected value' );
+
 	t.end();
 });
 
-tape( 'the function returns `NaN` if either argument is `NaN`', opts, function test( t ) {
+tape( 'the function returns `NaN` if either argument is `NaN` but not `+-infinity`', opts, function test( t ) {
 	var h;
 
 	h = hypot( NaN, 3.14 );