Core: Work around loss of precision from parseFloat

Fixes #15100

src/core.js

@@ -216,7 +216,8 @@ jQuery.extend({
 		// parseFloat NaNs numeric-cast false positives (null|true|false|"")
 		// ...but misinterprets leading-number strings, particularly hex literals ("0x...")
 		// subtraction forces infinities to NaN
-		return !jQuery.isArray( obj ) && obj - parseFloat( obj ) >= 0;
+		// adding 1 corrects loss of precision from parseFloat (#15100)
+		return !jQuery.isArray( obj ) && (obj - parseFloat( obj ) + 1) >= 0;
 	},
 
 	isEmptyObject: function( obj ) {

test/unit/core.js

@@ -462,7 +462,7 @@ test("isFunction", function() {
 });
 
 test( "isNumeric", function() {
-	expect( 37 );
+	expect( 38 );
 
 	var t = jQuery.isNumeric,
 		Traditionalists = /** @constructor */ function(n) {
@@ -490,6 +490,7 @@ test( "isNumeric", function() {
 	ok( t("4.536"), "Positive floating point string");
 	ok( t(-2.6), "Negative floating point number");
 	ok( t(3.1415), "Positive floating point number");
+	ok( t(1.5999999999999999), "Very precise floating point number" );
 	ok( t(8e5), "Exponential notation");
 	ok( t("123e-2"), "Exponential notation string");
 	ok( t(answer), "Custom .toString returning number");