Make rlp module more readable and organized with tests.

[Bhargavasomu]

- Also fixes potential bugs in rlp module
- Add custom tests and integrate with ethereum test fixtures

What was wrong?

The rlp module was closely structured to the yellowpaper making it hard to comprehend and maintain. Also there were a few potential bugs in the code which were cleaned up. One such code bug is as follows.

Decoding an empty bytestream should throw an error instead of returning bytearray([128])

Related to Issue #18

How was it fixed?

The code was rewritten using Encoder and Decoder classes to get clarity. The bugs have also been fixed.

Cute Animal Picture

[Bhargavasomu]

I am yet to add testcases for this module and also integrate test fixtures for this module. @lightclient @SamWilsn @adietrichs
before I do that, can you please do a preliminary review of the code?

[SamWilsn]

I'll take another look once the code is restructured :)

[Bhargavasomu]

@SamWilsn @lightclient @adietrichs I have added some personal tests and also integrated ethereum tests fixtures, and all of them seem to be working. Can you please review the code? Thanks.

[Bhargavasomu]

@SamWilsn @lightclient @adietrichs can you please review the code? Thanks.

[SamWilsn]

This is so much more readable! Great work. Just a bunch of me nitpicking really.

I did a few spot checks on the RLP test data, but we should probably make an issue to look into some differential fuzzing against other implementations (and/or steal their test suites.)

[lightclient]

Overall, nice job Somu! I left a few comments, but I think the big takeaways are:

• Where possible, name the magic RLP lengths - and when defining them, use hex notation. Almost everyone interacts with RLP in hex representation, so it's just natural to think of these things using their hex values.
• Avoid too much parameterization of tests and focus on readability instead.

[codecov-commenter]

Codecov Report

Merging #40 (e9b675f) into pyspec (632e395) will increase coverage by 8.31%.
The diff coverage is 98.94%.

@@            Coverage Diff             @@
##           pyspec      #40      +/-   ##
==========================================
+ Coverage   84.08%   92.40%   +8.31%     
==========================================
  Files          13       13              
  Lines         754      750       -4     
==========================================
+ Hits          634      693      +59     
+ Misses        120       57      -63     

Flag	Coverage Δ
unittests	92.40% <98.94%> (+8.31%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
src/eth1spec/rlp.py	99.02% <98.94%> (+58.84%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 632e395...e9b675f. Read the comment docs.

[SamWilsn]

I'm pretty happy with this, but I'd wait for @lightclient's review before merging.

[lightclient]

I'm sorry for being so nitpicky - I just think it's really important to use hex values as much we can when dealing with RLP since RLP is almost always viewed in that form. To most devs, numbers like 183, 192, etc. mean nothing to them. But things like 0xb7 and 0xc0 have meaning.