fix(pypi): show overridden index urls in pypi download error (#3130)

Closes #2985

Suppose invalid `experimental_index_url_overrides` in `pip.parse` is set
like below.

```bzl
pip.parse(
    experimental_index_url = "https://pypi.org/simple",
    experimental_index_url_overrides = {"mypy": "https://invalid.com"},
    hub_name = "pypi",
    requirements_lock = "//:requirements_lock.txt",
)
```

It fails as follows, showing only "pypi.org" as pypi index url, not
"invalid.com" for for `mypy` package.

```
Error in fail: Failed to download metadata for ["mypy"] for from urls: ["https://pypi.org/simple"].
If you would like to skip downloading metadata for these packages please add 'simpleapi_skip=["mypy"]' to your 'pip.parse' call.
```

To show overridden url for each package, show url of each package that
has been failed to download metadata. The error message with this PR is
like below.

```
Error in fail:
Failed to download metadata of the following packages from urls:
{
    "mypy": "https://invalid.com",
}

If you would like to skip downloading metadata for these packages please add 'simpleapi_skip=["mypy"]' to your 'pip.parse' call.
```