They're not the same.

When the encoded string's length modulu 4 is 2 or 3, there just need to be (at least) 2 or 1 padding characters ('=') for decoding to be successful, due to our decoder being rather strict. Less strict decoders may ignore the missing padding and successfully decode the encoded string.

When the remainder is 0, no padding is needed and everything is fine.

When the remainder is 1, the encoded string is simply invalid. It is not a padding issue. There is no valid way to decode the encoded string.