Revert "also disallow altsep beside pathsep inside paths"

[vphilippon]

This reverts commit 524d95e.

Fixes #1582
This essentially allows back the usage of / in pathname on Windows for virtualenvs for nested dirs.
Ex: python -m virtualenv foo/bar.

Thanks for contributing a pull request, see checklist all is good!

• wrote descriptive pull request text
• added/updated test(s)
• updated/extended the documentation
• added news fragment in docs/changelog folder

[gaborbernat]

This needs a test of using Unix paths on windows, and a changelog.

[vphilippon]

So I'm trying to understand how/where the path provided is used to write a proper test (rather than sticking the os.altsep component in some pre-existing matrix).

As I dig in, I'm getting an omnious feeling with how via_os_path.Path might behave in some cases.

I'll take some extra time to look this up. I'd like to make sure that the path get standardized with os.sep early and explicitly, rather than "hoping" that the mangling around eventually normalize it.

[gaborbernat]

With a test to validate that passing altsep paths work, I'm happy to go ahead with this. The test can be as simple as creating a env with activators and ensuring the paths within the activators use the sep instead of altsep.

[vphilippon]

In the end, since the creator is the piece providing, validating and sanitizing the dest path input, I've put a test on Creator.validate_dest. The activators currently rely on the creator giving a valid path, so I think this is the best way to test this.
Plus, that means there's no need to worry about os.altsep in other pieces past that point.

If you think of cases this won't cover let me know.