avoid reporting duplicate counterexamples

[karmacoma-eth]

There are a couple cases where this can happen:

• trivially, by adding the same assert multiple times:

    function test_duplicata1(uint256 x) external pure {
        uint256 mul = x * 42;
        assertNotEq(mul, 0);
        assertNotEq(mul, 0);
        assertNotEq(mul, 0);
    }

results in

Counterexample: 
    p_x_uint256_7e92c74_00 = 0x0000000000000000000000000000000000000000000000000000000000000000
Counterexample: 
    p_x_uint256_7e92c74_00 = 0x0000000000000000000000000000000000000000000000000000000000000000
Counterexample: 
    p_x_uint256_7e92c74_00 = 0x0000000000000000000000000000000000000000000000000000000000000000

• by introducing spurious paths before a single assert:

    function test_duplicata2(uint256 x) external pure {
        uint256 acc = 0;

        unchecked {
            // just introduce some extra paths
            acc += (x & (0xFF << 8) > 0) ? 2 : 3;
            acc += (x & (0xFF << 16) > 0) ? 5 : 3;

            assertNotEq((x & 0xff) * uint256(acc), 0);
        }
    }

• by having genuinely different asserts that result in the same cex values

    function test_duplicata3(uint256 x) external pure {
        assertNotEq(x, 0);
        assertNotEq(x & 0xff, 0);
        assertNotEq(x & 0xff << 8, 0);
    }

• ...

It seems a little odd and spammy to see the same cex repeated verbatim in the output.

Suggested behavior

• by default, only print the first counterexample (requires fixing --early-exit and making it the default)
• when not in early exit, make sure we only print unique cex values