[mypyc] Optimize str.encode with specializations for common used encodings

[svalentin]

Tested with:

import time
start = time.time()
for i in range(20000000):
    "test".encode('utf-8')
print(time.time() - start)

With PR applied and running mypyc, python3 -c "import test" runs in:
0.5383486747741699
0.5224344730377197
0.555696964263916

Without PR applied:
0.7315819263458252
0.7105758190155029
0.7471706867218018

Similar times observed for "ascii"

[JukkaL]

I think some cases aren't covered by the logic. Suggested some test cases that should help.

[JukkaL]

Also test cases where the specialization shouldn't be applied. Examples: s.encode(x), s.encode('a', x), s.encode('utf8', errors=x) and s.encode(errors=x) where x is not a literal.

Test cases where we have two keyword args: s.encode(encoding=..., errors=...) and s.encode(errors=..., encoding=...).

[svalentin]

I've updated the logic to work out the args better and added more tests. Please take another look!

[JukkaL]

Looks good, thanks!