Fix set type bit order

[threewholefish]

Python bistrings are processed MSB-first, but SBE set types are processed LSB first. Therefore, we need to reverse the encoding/decoding order as parsed.

This may need to be tested for sets which are larger than 1 byte

[kizzx2]

This changes how the parser works. Since I and other users have been using this for real world data and apparently those worked, can you provide some example data of why this change is needed?

[threewholefish]

Take this set from the CME MDP3 spec:

<set name="MatchEventIndicator" encodingType="uInt8">                                                  
    <choice name="LastTradeMsg" description="1=Last trade message for the event, 0=Not last">0</choice>
    <choice name="LastVolumeMsg" description="1=Last electronic volume message, 0=Not last">1</choice>
    <choice name="LastQuoteMsg" description="1=Last real quote message, 0=Not last">2</choice>
    <choice name="LastStatsMsg" description="1=Last statistics message, 0=Not last">3</choice>
    <choice name="LastImpliedMsg" description="1=Last implied quote message, 0=Not last">4</choice>
    <choice name="RecoveryMsg" description="1=Message is sent during  recovery process">5</choice>
    <choice name="Reserved" description="0=Reserved for future use">6</choice>
    <choice name="EndOfEvent" description="1=Last message for the event, 0=Not last">7</choice>
</set>

The raw value 0x1 should correspond to a decoded value of ["LastTradeMsg"], but it is instead decoded as ["EndOfEvent"]. Encoding is consistent. As best I can tell, this is because the SBE spec states that the bit indices start from the LSB, but bitstring.Bits and bitstring.Bitarray start from the MSB.