fix handling of env-vars passed to plumbum Commands; support new with_cwd

[koreno]

Fix handling of env-vars passed to plumbum BoundEnvCommands:
- don't use the non-threadsafe and session-dependent .env() context manager
- sync popen support for 'env' param in all machine impls: local/ssh/paramiko
- add .with_cwd() to complement .with_env()

[coveralls]

Coverage increased (+0.0002%) to 82.293% when pulling e4db361 on fix-cmd-with-env into f817e62 on master.

[AndydeCleyre]

Is there more documentation for with_cwd? How does it relate to the usual local.cwd? Would the following be valid?

from plumbum import local
from plumbum.cmd import pwd

with local.cwd('/tmp'):
    pwd.with_cwd('/home')().strip() == '/home'
    pwd().strip() == '/tmp'

[henryiii]

Could you please address @AndydeCleyre's question? I'm a little worried if we are just adding another way to spell something that already exists, that's just more to maintain and adapt for Plumbum 2.0.

[koreno]

Is there more documentation for with_cwd? How does it relate to the usual local.cwd? Would the following be valid?

from plumbum import local
from plumbum.cmd import pwd

with local.cwd('/tmp'):
    pwd.with_cwd('/home')().strip() == '/home'
    pwd().strip() == '/tmp'

sorry it took so long -
I would expect the with_cwd to override the ctxmgr, since it is set directly on the command object. Only command objects that don't have their own cwd will fall back on the default, which is modified by the ctxmgr.
I will add a test for this, and docs.

What do you think?

[henryiii]

With test and docs, I'd be fine with it.

[koreno]

@henryiii - ci is failing on coverall actions, not sure how to proceed...

[henryiii]

Looks like it is due to this: coverallsapp/github-action#7

We'll probably have to disable coverage on PRs that are forked, unless there's some other way around this. We can print a report, but we can't upload it to coveralls. pull_request_target is a new option, but it runs on the base and not the merged state, so can't be used for coverage.

[henryiii]

Oh, that's odd, this is not a fork. Should be getting a read/write token. (Probably still need to fix this for forks, but not as sure this is the problem).

[henryiii]

Coveralls 3 broke GitHub service detection ( TheKevJames/coveralls-python#252 ), so I now manually specify it and it should be working again.