Calculator: Add KeypadValue class

This commit adds a basic KeypadValue class which abstracts away
Keypad's internal representation in a slightly simpler format.
This will allow arithmetic operations in the Calculator to not
lose any precision. In cases where losing precision is necessary,
an explicit conversion operator to double is provided, as well as
an explicit constructor from double.

Userland/Applications/Calculator/KeypadValue.h

@@ -0,0 +1,42 @@
+/*
+ * Copyright (c) 2021, the SerenityOS developers.
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#pragma once
+
+#include <AK/Math.h>
+#include <AK/String.h>
+#include <AK/Types.h>
+
+class KeypadValue {
+    friend class Keypad;
+
+public:
+    KeypadValue(i64, u8);
+    KeypadValue(i64);
+
+    KeypadValue operator+(KeypadValue const&);
+    KeypadValue operator-(KeypadValue const&);
+    KeypadValue operator*(KeypadValue const&);
+    KeypadValue operator-(void) const;
+    bool operator<(KeypadValue const&);
+    bool operator>(KeypadValue const&);
+    bool operator==(KeypadValue const&);
+
+    explicit KeypadValue(double);
+    explicit operator double();
+
+private:
+    template<typename T, typename F>
+    T operator_helper(KeypadValue const& lhs, KeypadValue const& rhs, F callback);
+
+    // This class represents a pair of a value together with the amount of decimal places that value is offset by.
+    // For example, if we were to represent the value -123.55 in this format, m_value would be -12355 and
+    // m_decimal_places would be 2, because when you shift -12355 2 digits to the right, you get -123.55.
+    // This way, most operations don't have to be performed on doubles, but can be performed without loss of
+    // precision on this class.
+    i64 m_value { 0 };
+    u8 m_decimal_places { 0 };
+};