Add support for PEP-646

[zsol]

Summary

Implementation and tests based on python/cpython#31018
The CST is modeled slightly differently from the AST:

• Index can now have an optional star (and corresponding whitespace_after_star)
• StarredElement is now a BaseExpression

Test Plan

• roundtrip tests
• negative parse tests
• parse tests
• node construction tests

TODO

• Maybe make whitespace_after_star into a MaybeSentinel instead of Optional and add corresponding codegen logic

[codecov-commenter]

Codecov Report

Merging #696 (0130225) into main (380f045) will decrease coverage by 0.02%.
The diff coverage is 82.05%.

@@            Coverage Diff             @@
##             main     #696      +/-   ##
==========================================
- Coverage   94.82%   94.79%   -0.03%     
==========================================
  Files         246      246              
  Lines       25653    25689      +36     
==========================================
+ Hits        24325    24353      +28     
- Misses       1328     1336       +8     

Impacted Files	Coverage Δ
libcst/_nodes/tests/base.py	98.29% <ø> (ø)
libcst/matchers/_return_types.py	100.00% <ø> (ø)
libcst/_nodes/tests/test_funcdef.py	86.66% <78.57%> (-6.89%)	⬇️
libcst/_nodes/expression.py	96.66% <80.00%> (-0.13%)	⬇️
libcst/_typed_visitor.py	97.00% <83.33%> (-0.04%)	⬇️
libcst/matchers/__init__.py	100.00% <100.00%> (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 380f045...0130225. Read the comment docs.

[tusharsadhwani]

Can whitespace_after_star not be a MaybeSentinel then? Are there complications?

[zsol]

@tusharsadhwani I decided against it because there doesn't seem to be any difference in behavior (the value of None always translates to the same as MaybeSentinel would) and Optional seems more semantically correct. I'm open to be convinced otherwise, though