Function return type depends on previous calls

Nobody has claimed this yet.

Function type inference changes based on previous calls of the same function. [Playground example](https://pyright-play.net/?code=MYGwhgzhAEBKCmAHAlmALvATgLgFDQOgBN4AzaeYACwHsBtAFQF0AKCeEUgGmgA9toDAJTQAtAD5BeQjOiZ4aAK6YAdn1y4S5Ui148AnkOmF5S1XwB0lWi0MbM0ALxwkqDJhZDcYJ9B2YeAEYRWQIAYmgBZBU0XAAjX38eACIaFXhkkPDI6Gi0aAAfaAg0AOg4xXzkckR5ADdkGkUYEGj4XJh5AFsaOvgiHJLMIA)
```python
class Repiater:
def echo[T](self, x: T) -> T:
return x

def f(x, y):
return x.echo(y)

r = Repiater()
a = f(r, 1) # : int
b = f(r, "one") # : int | str, but if previous line is removed : str
```
The inferred type of `f`’s return value depends on whether earlier calls exist. Not necessarily a bug, but unexpected behavior.

Start with the linked Playground example and reproduce the inferred return types for f with and without the earlier call. Trace how Repiater.echo and f are inferred, then determine whether the result is expected or a bug; done means the behavior is explained and any required change has a regression test.

Written by the indexing model from the issue text.