100 Percent test coverage means nothing

FullCoverage.cs

using System;
using Microsoft.VisualStudio.TestTools.UnitTesting;

namespace TDDCoverage
{
    public enum Gender { Unknown, Male, Female }

    public class Customer
    {
        public Gender Gender = Gender.Unknown;
        public bool IsVIP = false;
        public bool IsPreferred = false;
    }

    public class Discounter
    {
        // A customer can get a discount based on it's profile.
        // Vips:
        // - A male gets a 44% discount
        // - A female gets a 33% discount
        // - If we don't know the gender we get a 22% discount
        // Preferred customers:
        // - A preferred customer gets a discount of 10%
        public static decimal DiscountPercentage(Customer c)
        {
            decimal discount = 0;
            if (c.IsVIP)
            {
                if (c.Gender == Gender.Male) discount = 44;
                else if (c.Gender == Gender.Female) discount = 33;
                else discount = 22;
            }
            if (c.IsPreferred)
            {
                discount = 10;
            }
            return discount;
        }
    }


    // This testcase covers all the code, but the code still has a bug
    [TestClass]
    public class FullCoverage
    {
        [TestMethod]
        public void Discount_regular()
        {
            Assert.AreEqual(0, Discounter.DiscountPercentage(new Customer { }));
        }

        [TestMethod]
        public void Discount_vip()
        {
            Assert.AreEqual(44, Discounter.DiscountPercentage(new Customer { Gender = Gender.Male, IsVIP = true }));
            Assert.AreEqual(33, Discounter.DiscountPercentage(new Customer { Gender = Gender.Female, IsVIP = true }));
            Assert.AreEqual(22, Discounter.DiscountPercentage(new Customer { Gender = Gender.Unknown, IsVIP = true }));
        }

        [TestMethod]
        public void Discount_Preferred()
        {
            Assert.AreEqual(10, Discounter.DiscountPercentage(new Customer { IsPreferred = true }));
        }

        // The problem with the code is that we actually overrule the vip 
        // behaviour with the preferred behaviour, which is obviously a bug.
        // 
        // While it might be obvious here, it might not be as obvious 
        // in the big codebases one usually works with, as the cyclomatic 
        // complexity of these is WAAAY higher then this case.
        //
        // The point I want to make is that 100% test coverage proves nothing,
        // because you only test specific scenarios, and not all the possible 
        // combinations. Most bugs in software appear because the software was 
        // used in an unexpected way.
        //
        // If you would like to test all behaviour, one has to fall back to 
        // property based testing.
        //
        // This is the missing test case:
        //
        //[TestMethod]
        //public void Discount_preferred_and_vip()
        //{
        //    Assert.AreEqual(44, Discounter.DiscountPercentage(new Customer { Gender = Gender.Male, IsVIP = true, IsPreferred = true }));
        //    Assert.AreEqual(33, Discounter.DiscountPercentage(new Customer { Gender = Gender.Female, IsVIP = true, IsPreferred = true }));
        //    Assert.AreEqual(22, Discounter.DiscountPercentage(new Customer { Gender = Gender.Unknown, IsVIP = true, IsPreferred = true }));
        //}
    }
}