Reverse trie key bit ordering

[arkpar]

As per changes in GP 0.4.0

[sourabhniyogi]

Thanks for this -- We are able to pass this.

[vekexasia]

The test files included in this pr are not matching what is written in graypaper.

Leaf with |v|<=32 is encoded wrongly by performing a bitwise or of [1,0] and the length of |v|.

in case |v| is = 1 (for example) its binary repr is 00 00 00 01 . Graypaper says we should use the first 6 which corresponds to 00 00 00 but the tests here do a different thing.

vs

 if len(v) <= 32:
        head = 0b10000000 | len(v)
        return bytes([head]) + k[:-1] + v + ((32 - len(v)) * b'\0')

TBH the implementation here makes more sense than taking the first 6 digits considering |v|<=32.

I just wanted to write it down so that we know its currently out of spec (0.5.2) but may become the correct tests in a future GP release.

[arkpar]

@vekexasia Thank you for noticing this. Should be fixed in the spec:
gavofyork/graypaper#168