Merge pull request #449 from numpy/docs/fix-extended-precision

📝 fix incorrectly documented extended-precision removals

Author: jorenham

docs/user_guide/differences.md

@@ -83,7 +83,9 @@ The following non-existent scalar types have been removed (numpy/numtype#209):
 - `int128` and `int256`
 - `uint128` and `uint256`
 - `float80` and `float256`
-- `complex128` and `complex256`
+- `complex160` and `complex512`
+
+These types will not be defined on any supported platform.
 
 ## Aliases of `longdouble`