Reorder functions within the bytecode

[iamdefinitelyahuman]

What I did

Reorder functions within the bytecode and refactor how the function selector is checked.

How I did it

Functions were previous sequenced in the same way as they are found in the source. Internal functions were jumped over following an inefficient PUSH1 0x0 ISZEZO check.

This PR places all internal functions after the fallback, so they cannot be reached without being explicitely jumped to from an internal function. Thus, all the jumping logic can be removed and the bytecode size reduced.

The function selector itself was previous written to memory with offset 0, and loaded onto the stack for every comparison. This PR modifies the process so that the selector is kept on the stack and compared via DUP operations rather than repeated MLOAD. Less bytecode, less gas.

How to verify it

Run the tests, see that nothing has broken.

Cute Animal Picture

Coming soon

[codecov-io]

Codecov Report

Merging #2303 (c22c081) into master (fbd6880) will decrease coverage by 0.03%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master    #2303      +/-   ##
==========================================
- Coverage   85.82%   85.79%   -0.04%     
==========================================
  Files          85       85              
  Lines        8546     8549       +3     
  Branches     2036     2039       +3     
==========================================
  Hits         7335     7335              
- Misses        728      730       +2     
- Partials      483      484       +1     

Impacted Files	Coverage Δ
vyper/compiler/utils.py	96.42% <100.00%> (+0.13%)	⬆️
...er/function_definitions/parse_external_function.py	100.00% <100.00%> (ø)
...er/function_definitions/parse_internal_function.py	97.89% <100.00%> (-0.05%)	⬇️
vyper/parser/function_definitions/utils.py	100.00% <100.00%> (ø)
vyper/parser/parser.py	81.81% <100.00%> (+0.64%)	⬆️
vyper/functions/functions.py	89.25% <0.00%> (-0.40%)	⬇️

---

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 fbd6880...095d5f8. Read the comment docs.