Issue equating zero `big.{Int,Float}`'s

[lmittmann]

When comparing std big.Int's or big.Float's there is an edge case for the value zero where is reports the error 0 != 0. The reason for this is the big.{Int,Float} attribute neg bool which can be both true and false for the value zero.

Example

package main

import (
	"math/big"
	"testing"

	"github.com/matryer/is"
)

func TestBig(t *testing.T) {
	is := is.NewRelaxed(t)
	big0 := new(big.Int).Add(big.NewInt(-1), big.NewInt(1))
	is.Equal(new(big.Int), big0)              // big.Int
	is.Equal(new(big.Float), big.NewFloat(0)) // big.Float
}

This will result in:

        is_test.go:13: 0 != 0 // big.Int
        is_test.go:14: 0 != 0 // big.Float
--- FAIL: TestBig (0.00s)
FAIL

It's not really a bug since the structs are different after all. However, since this are std types it would be nice if they would be compared by their Cmp function.

[lmittmann]

@matryer would you accept a PR?

[breml]

Hi @lmittmann
There was a longer discussion about is.Equal and some short comings it currently has in #53 . Before providing a PR, I suggest to you go through the discussion there.