Highlight differences in complex types

[JukkaL]

Sometimes it's hard to see why two types are incompatible if they are long and almost the same. Dummy example:

from typing import Callable
from mypy_extensions import Arg

x: Callable[[Arg(int, 'foobar'), Arg(int, 'bir'), Arg(int, 'fubar')], None]
y: Callable[
    [Arg(int, 'foobar'), Arg(int, 'blr'), Arg(int, 'fubar')], None]

x = y

This is the error:

t.py:7: error: Incompatible types in assignment (expression has type "Callable[[Arg(int, 'foobar'), Arg(int, 'blr'), Arg(int, 'fubar')], None]", variable has type "Callable[[Arg(int, 'foobar'), Arg(int, 'bir'), Arg(int, 'fubar')], None]")

Here's a potential better message that highlights the difference:

t.py:7: error: Incompatible types in assignment (expression has type "Callable[[Arg(int, 'foobar'), Arg(int, 'blr'), Arg(int, 'fubar')], None]", variable has type "Callable[[Arg(int, 'foobar'), Arg(int, 'bir'), Arg(int, 'fubar')], None]")
t.py:7: note: Expression type: ..., Arg(int, 'blr'), ...
t.py:7: note: Variable type:   ..., Arg(int, 'bir'), ...
t.py:7: note:                                 ^^^

It's not clear what's the best way to generalize this sort of "type diff" to arbitrary nested types and to arbitrary type differences. We should at least handle some typical cases. We could also look at how other (non-Python) language implementations handle this.

Just splitting long lines and aligning them would help a bit (but in general the alignment could be off, since there could be multiple differences in types, only some of which are problematic):

t.py:7: error: Incompatible types in assignment (expression has type "Callable[[Arg(int, 'foobar'), Arg(int, 'blr'), Arg(int, 'fubar')], None]", 
t.py:7: error:                                     variable has type "Callable[[Arg(int, 'foobar'), Arg(int, 'bir'), Arg(int, 'fubar')], None]")