Fix memory length generated by assert_reason

[ZumZoom]

Fixes #1981

Description of issue

Code generation for assert_reason generates incorrect length for revert string which leads to issues with revert reason decoding.

How I fixed it

Fixed length of memory passed to revert opcode generated by assert_reason.

How to verify it

Check that assert tests are still passing.

Description for the changelog

Removed 32 bytes of junk at the end of revert reasons.

Cute Animal Picture

[iamdefinitelyahuman]

This looks right to me. A followup to my comments in #1981, I think we actually do need the PUSH32 to ensure the memory is zero'd.

[codecov-io]

Codecov Report

Merging #1982 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master    #1982   +/-   ##
=======================================
  Coverage   87.17%   87.17%           
=======================================
  Files          58       58           
  Lines        6637     6637           
  Branches     1677     1677           
=======================================
  Hits         5786     5786           
  Misses        536      536           
  Partials      315      315           

Impacted Files	Coverage Δ
vyper/parser/stmt.py	92.40% <ø> (ø)

---

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 1ebce95...f87aaf6. Read the comment docs.