Remove Maximum Table Size test

[314eter]

As discussed in #78, section 4.2 can be interpreted in different ways. But even if you interpret it in the most restrictive way (all table size updates MUST occur at the beginning of a header block), it's only a MUST for the encoder. It says nowhere that the decoder "MUST treat this as a decoding error", which is what the test is checking.

So I don't think there's anything wrong with a decoder that accepts table size updates in the middle of a header block, because it's still perfectly compatible with encoders that follow the other interpretation. For encoders, it's probably safer to be more strict in order to be compatible with all decoders.