enable infix_spaces_linter, and apply it to the package

[MichaelChirico]

Closes #594

[AshesITR]

Codecov complains about uncovered code in object_name_linters.R (seems unused - remove?) and namespace.R (haven't looked into it)

[MichaelChirico]

I think the namespace one is actually a legit bug. namespace_imports is only called here:

https://github.com/jimhester/lintr/blob/fa05ceb2bf967ddad9230e25879b95084c80b0dc/R/object_name_linters.R#L57

without an argument. the default is to run find_package() (again without argument), but there's no default, so normalizePath always fails:

https://github.com/jimhester/lintr/blob/fa05ceb2bf967ddad9230e25879b95084c80b0dc/R/lint.R#L287

In namespace_imports this error happens within tryCatch so it just returns NULL.

I'm not sure the intended behavior but certainly the above is not it 😃

[MichaelChirico]

For the unused code in object_name_linters.R, I removed that code in #621, so let's just merge that one first.

Hopefully then we'll pass codecov here.

[AshesITR]

The intended behaviour seems to be to find_package() on the to-be-linted file, i.e. namespace_imports() needs the filename of the linted file, check for a DESCRIPTION and extract imports from other packages based on that information.
I'll create a follow-up issue from this.

[AshesITR]

Thank you :)

[AshesITR]

R-devel apparently changed the error message for the lint we use in test-error.R.
I'll create a PR to allow the new error message ("unexpected end of line") as well as the old one ("unexpected input")

[AshesITR]

Blocked by #659

[MichaelChirico]

The coverage miss here is caused by #642 -- I touched a region of code that's not reachable currently. Merging anyway, that should be tested as part of the fix to that issue.