Do not set extra in default_environment()

[sbidoul]

PEP 508 says explicitly that the presence of the extra key in an environment is an error except in special cases.

This PR tweaks #550. cc/ @MrMino @uranusjr @pradyunsg

[pradyunsg]

The specific section:

The “extra” variable is special. It is used by wheels to signal which specifications apply to a given extra in the wheel METADATA file, but since the METADATA file is based on a draft version of PEP 426, there is no current specification for this. Regardless, outside of a context where this special handling is taking place, the “extra” variable should result in an error like all other unknown variables.

In other words, the extra key is only for dependency metadata evaluation, which is the only context that markers get evaluated in?

[sbidoul]

In other words, the extra key is only for dependency metadata evaluation, which is the only context that markers get evaluated in?

Yes exactly. That is why I think it is better to set it in evaluate() rather than default_environment() which could be used in other contexts than marker evaluation (as in, "just give me the environment of the running interpreter").

Also, #550 did change the expected result in the test of default_environment() (which I revert here). I think that might cause issues downstream and that was not necessary to achieve the goal of that PR.